A Cross-Cultural Comparison

RESEARCH REPOSITORY

This is the author’s final version of the work, as accepted for publication following peer review but without the publisher’s layout or pagination. The definitive version is available at:

https://doi.org/10.1016/j.im.2018.05.006

Liu, Z. and Wang, X. (2018) How to regulate individuals’ privacy boundaries on social network sites: A Cross-Cultural comparison. Information & Management

http://researchrepository.murdoch.edu.au/id/eprint/40902/

Accepted Manuscript

Title: How to Regulate Individuals’ Privacy Boundaries on Social Network Sites: A Cross-Cultural Comparison

Authors: Zilong Liu, Xuequn Wang

PII: S0378-7206(17)30277-X DOI: https://doi.org/10.1016/j.im.2018.05.006 Reference: INFMAN 3073

To appear in: INFMAN

Received date: 31-3-2017 Revised date: 2-5-2018 Accepted date: 4-5-2018

Please cite this article as: Zilong Liu, Xuequn Wang, How to Regulate Individuals’ Privacy Boundaries on Social Network Sites: A Cross-Cultural Comparison, Information and Management https://doi.org/10.1016/j.im.2018.05.006

This is a PDF ﬁle of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its ﬁnal form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. How to Regulate Individuals’ Privacy Boundaries on Social Network Sites: A Cross-Cultural Comparison Zilong Liu School of Management Science and Engineering, Dongbei University of Finance and Economics, Dalian, Liaoning, China Email: [email protected] Xuequn Wang (Corresponding Author) School of Engineering and Information Technology Murdoch University, Perth, WA, Australia, 6150 Phone: +61 8 93602793 [email protected] Funding: This work is supported by the grant of National Natural Science Foundation of China (Nos. 71771040; 71301021) and internal grant from School of Engineering and Information Technology, Murdoch University. How Individuals Regulate Their Privacy Boundaries on Social Networking Sites: A Cross-

Cultural Comparison

Abstract Individuals presently interact with their diverse social circles on social networking sites and may find it challenging to maintain their privacy while deriving pleasure through self- disclosure. Drawing upon the communication privacy management theory, our study examines how boundary coordination and boundary turbulence can influence individuals’ self-disclosure decisions. Further, our study examines how the effects of boundary coordination and boundary turbulence differ across cultures. Our hypotheses are tested with survey data collected from the United States and China. The results strongly support our hypotheses and show interesting cultural differences. The implications for theory and practice are discussed. Keywords: social networking sites; privacy; communication privacy management; role conflict; self-disclosure; culture

ACCEPTED MANUSCRIPT

Introduction

Presently, individuals often interact with their diverse social circles (i.e., relatives, schoolmates, teachers, colleagues, friends, online friends, etc.) on social networking sites

(SNSs). In such a context, they may find meeting the inconsistent expectations of their diverse social circles quite challenging, if not impossible (Besmer and Lipford, 2010). As a result, individuals’ privacy boundaries become blurred, and they are vulnerable to potential privacy threats (e.g., misuse of disclosed information) (Brandtzæg et al., 2010). For example, pictures may be taken when individuals have parties and drink alcohol with their friends. The friends may expect to see the pictures posted on Facebook to show their friendship. However, the individuals may be in sports teams and worry that their coaches can also view the pictures

(Besmer and Lipford, 2010).

Indeed, individuals presently are increasingly integrating the use of SNSs into their lives.

A recent report shows that in 2016, Internet users spent 118 minutes on average per day using

SNSs (Statista.com, 2016). On SNSs such as Facebook, Twitter, and WeChat (a Chinese mobile SNS), individuals maintain their existing social relationships and develop new relationships by disclosing their personal information (Boyd and Ellison, 2008; Wang and Li,

2015). They can update their status, upload their pictures, and comment on information disclosed by others. From the perspective of the SNSs, supporting individuals’ information disclosure is vital because their market values depend heavily on the social interactions among their users (Smith, 2013). However, while individuals enjoy the pleasure and gratification derived from interacting with their contacts on SNSs, their privacy is endangered (ZhangACCEPTED et al., 2011). In particular, Facebook usersMANUSCRIPT now have approximately 228 contacts on average (Mazin, 2014). On WeChat, the average number of contacts is around 128 per user

(WeChat, 2015). As the number of contacts on SNSs increases, individuals realize that they need to interact with their diverse social circles. However, as illustrated above, individuals

may have inconsistent expectations and experience conflicts from these social circles. To avoid potential conflicts, individuals may refuse to upload pictures or even break certain social connections, which will ultimately hurt the value of the SNS. Therefore, it is important for SNSs to understand how to protect individuals’ privacy and support their self-disclosure in such scenarios.

Previous literature has examined privacy in various contexts (Bélanger and Crossler, 2011;

Liu et al., 2016; Smith et al., 2011), and recent literature has begun to pay more attention to privacy on SNSs. Our review of the previous literature (Table A.1 in Appendix A) shows that most studies focus on how individuals manage their privacy and make self-disclosure decisions themselves. However, on SNSs, once information is disclosed, it becomes accessible to both senders (i.e., those who disclose the information) and receivers (i.e., others on SNSs). Receivers here can come from individuals’ diverse social circles. Both sides are responsible for controlling the disclosed information, and the fact that these diverse social circles co-exist on SNSs has important implications for individuals’ self-disclosure. However, few studies have examined self-disclosure from a collective privacy management perceptive and assessed the influence of individuals’ various social circles. Such studies are highly needed because collective privacy management differs from individual privacy management and involves interpersonal interactions and perception regarding how disclosed information will be co-managed (Xu, 2012). Further, as discussed above, conflicts that individuals experience while interacting with diverse social circles can impede their social connections and hurt the value of SNSs. Therefore, examining self-disclosure from a collective privacy managementACCEPTED perspective is vital because it can MANUSCRIPT provide better understanding of how to support individuals’ social connections, and also, it can provide information about SNS providers’ operational strategies. One exception is De Wolf et al. (2014), who examined both individual and group privacy management. However, their study focuses on members of a

youth organization on Facebook where only one social circle (i.e., other members of the youth organization) is involved in social interaction1.

Therefore, to better understand individuals’ information disclosure and support their social interactions on SNSs, our first objective is to examine how SNS users form their privacy boundaries collectively with others and make their information disclosure decisions while interacting with different social circles on SNSs. In other words, we focus on how individuals’ different social circles affect their self-disclosure on SNSs, which is not well examined in the previous literature. Our study draws upon Petronio’s (2002) communication privacy management (CPM) theory to understand how individuals form their privacy boundaries and make self-disclosure decisions. In other words, our study does not focus on individuals’ privacy boundaries per se but on how their privacy boundaries are formed while interacting with different social circles on SNSs. Previous literature has applied the CPM theory to understand how institutional privacy assurances can reduce individuals’ privacy concerns (Xu et al., 2011). Our study extends the previous literature by applying the CPM theory in the context of SNSs and proposing corresponding mechanisms for boundary coordination and turbulence.

Further, SNS providers presently often operate in different countries and deal with users from diverse cultural backgrounds, and it is necessary to understand how users form their privacy boundaries across culture. Culture is thus an important factor influencing SNS users’ self-disclosure decisions (Cullen, 2009; Miltgen and Peyrat-Guillard, 2014). However, our review (Table A.2 in Appendix A) shows that culture has not been well examined in the previousACCEPTED cross-cultural privacy literature, even MANUSCRIPT though it is recognized in the CPM theory

1 In another qualitative study, Brandtzæg et al. (2010) argue that a mix of different social circles can influence individuals’ self-disclosure on SNSs. However, their study does not clarify theoretical mechanisms through which diverse social circles influence self-disclosure.

(more details in “Literature Review and Theory Foundation” section). Many studies treat culture as given and assume that certain cultural dimensions are different when comparing two countries. Some other studies collect certain cultural dimensions and use them as independent variables. These studies can be helpful to understand how cultural dimensions can influence individuals’ privacy concerns but cannot help assess how certain relationships are different under different cultural dimensions. On the other hand, few studies try to understand individuals’ privacy under different espoused cultural dimensions.

To further understand the effect of culture on individuals’ privacy and self-disclosure decisions, we draw upon the CPM theory and Hofstede’s national cultural dimensions

(Hofstede, 2001). Thus, our second objective is to examine the moderating role of culture in individuals’ forming their privacy boundaries on SNSs while interacting with different social circles on SNSs. The United States (US) and China are selected as exemplars of distinct cultural differences, and we collect different cultural dimensions from Hofstede (2001). By comparing SNS users in the US and China as well as in subsamples of different espoused cultural dimensions, our study can clarify how individuals from different cultures perceive privacy and form their privacy boundaries.

To summarize, our study examines how SNS users form their privacy boundaries and make self-disclosure decisions while interacting with different social circles on SNSs, as well as the moderating effect of culture on the process of privacy boundary formation. Our study focuses on individuals’ disclosing their personal information rather than all types of information sharing (e.g., news sharing) on SNSs. Our study makes two important contributions.ACCEPTED First, our study contributes to theMANUSCRIPT privacy literature by clarifying how individuals make their privacy decisions while interacting with various social circles on

SNSs, an issue that has not been well examined in the literature. Our study also extends the previous literature applying the CPM theory by clarifying the mechanisms of boundary

coordination and turbulence in the context of SNSs. The results of our study can provide information about SNS providers’ operational strategies and support users’ social interactions on SNSs. Second, our study clarifies the role of culture in the process of self-disclosure. The findings can provide valuable insights for SNS providers regarding how to protect individuals’ privacy across different cultures.

The rest of our paper is organized as follows. We first briefly review the previous literature and discuss our research objectives to fill the gaps in the literature. Next, we introduce our theoretical background and discuss how culture can be relevant to understanding privacy. We then present our research model and develop our hypotheses.

Thereafter, research methodology is presented, and the results of the data analysis are reported. Finally, our paper concludes with implications for theory and practice, the limitations of our study, and opportunities for future studies.

Literature Review and Theory Foundation

Privacy Literature: A Brief Review

As SNSs become part of individuals’ lives, practitioners are interested in how to support individuals’ information disclosure on SNSs. Individuals’ information disclosure is critical because it can not only support their social interactions with other users but also greatly influence market values of SNSs (Smith, 2013). Previous privacy and self-disclosure literature in the context of SNSs has examined different factors influencing individuals’ information disclosure. Our review (Table A.1 in Appendix A) shows that these factors can be divided into four categories. The first category is individual attributes including extroversionACCEPTED (Chen, 2013) and information disclosure MANUSCRIPT self-efficacy (James et al., in press). Factors from this category show that individuals with different attributes have different levels of self-disclosure intention. The second category is social interaction attributes including social identity (Shih et al., 2017), self-expression (James et al., 2015), and relationship

building/maintenance (e.g., Krasnova et al., 2010). Factors from this category show how self- disclosure supports individuals’ social interactions with others on SNSs. For example, self- disclosure can help individuals build social relationships with others. Therefore, relationship building has a positive effect on self-disclosure (Krasnova et al., 2010). The third category is service attributes including perceived critical mass (Chen, 2013) and privacy policy (Gerlach et al., 2015). Factors from this category focus on services provided by SNSs. For example, individuals want to see that SNSs have strong privacy policies to ensure that their disclosed information can be well protected. The last category is external environment attributes including perceived Internet risk (Chen, 2013). Factors from this category reflect how individuals perceived the Internet environment. When individuals disclose information on

SNSs, the information needs to travel through the Internet. Therefore, a safe Internet environment is also needed to facilitate self-disclosure.

Although the pervious literature has provided valuable insights to advance our understanding of self-disclosure on SNSs, these studies mainly focus on how individuals make self-disclosure decisions themselves and few studies have examined the role of other users from their different social circles in the process of self-disclosure. On SNSs, disclosed information can be accessed by individuals’ contacts, who may come from different social circles. As individuals’ social circles become more diverse, they may experience conflicts and their privacy boundaries become blurred. They may thus feel confused regarding what information should be disclosed and how the disclosed information will be treated. In certain scenarios, such conflicts may result in negative impacts on individuals. In the example discussedACCEPTED above, if individuals in sports teams MANUSCRIPT choose to upload their party pictures, they may lose their positions in teams and their professional careers may suffer. Therefore, it is important to understand how individuals’ diverse social circles influence their self-disclosure on SNSs, which has not received much attention from previous literature (Liu et al., 2018b).

To fill this gap, our first research objective is to examine how SNS users form their privacy boundaries and make their information disclosure decisions when interacting with different social circles on SNSs.

Further, SNS users may come from diverse cultural backgrounds, and it is necessary to understand how they form their privacy boundaries across culture. Previous literature has suggested that culture is an important factor influencing SNS users’ self-disclosure decisions

(Cullen, 2009; Liu et al., 2018a; Miltgen and Peyrat-Guillard, 2014). Our review (Table A.2 in Appendix A) shows that previous privacy literature has mainly dealt with culture following two approaches. The first approach treats culture as given (Myers and Tan, 2002) and assumes that certain cultural dimensions are different when comparing two countries. For example, Americans are regarded to have a low level of collectivism, whereas Chinese are regarded to have a high level of collectivism (Hofstede, 2001). Then the relationships between social rewards and self-disclosure may be tested between American and Chinese, assuming that the differences in these relationships are caused by different levels of individualism. Such studies can be helpful to understand how the relationships are different between Americans and Chinese, but the differences may or may not be attributed to the various levels of individualism. It is possible that the differences are due to other culture dimensions or other contextual factors.

The second approach uses certain cultural dimensions as independent variables. For example, Trepte et al. (2017) find that individualism has a direct effect on willingness for uploading pictures. These studies can be helpful to understand how cultural dimensions can influenceACCEPTED individuals’ privacy concerns but cannotMANUSCRIPT help assess how certain relationships are different under various cultural dimensions.

Therefore, based on our review, few studies try to examine individuals’ privacy under different espoused cultural dimensions. Espoused cultural values refer to “the degree to

which an individual embraces the values of his or her national culture.” (Srite and Karahanna,

2006, p. 681). In other words, few studies have measured individuals’ different espoused cultural dimensions and assessed espoused culture dimensions’ moderating effects on individuals’ privacy perceptions and self-disclosure decisions. Such studies are highly important to advance our understanding of how individuals’ self-disclosure decisions are influenced by culture dimensions. For example, consider the context where researchers want to examine how collectivism moderates the effect of social rewards on intention to self- disclose. Without measuring espoused cultural dimensions and assessing their moderating effects, it is not possible to understand the effect of social rewards under different levels of collectivism.

To fill this gap, our second research objective is to examine the moderating role of culture in individuals’ forming their privacy boundaries on SNSs. Specifically, we treat espoused culture dimensions as moderators and examine moderating effects of espoused culture dimensions on individuals’ forming their privacy boundaries. In the following, we introduce our theoretical foundations: the CPM theory and national culture dimensions. Here, the CPM theory is helpful to understand how individuals form their privacy boundaries and make self-disclosure decisions, whereas national culture dimensions are useful to clarify the moderating effect of different culture dimensions.

Privacy Boundary Management

Privacy concerns can influence important outcomes such as intention to transact (Dinev and Hart, 2005) and self-disclosure (Chellappa and Sin, 2005). Consistent with Xu et al. (2011),ACCEPTED our study focuses on situation-specific MANUSCRIPT privacy concerns, defined as SNS users’ concerns about possible loss of privacy as a result of information disclosure to someone unknown.

There has been a debate about the definition of privacy (i.e., what is privacy?). As summarized in Smith et al. (2011), the previous literature has followed two broad approaches to define privacy: the value-based approach or the cognate-based approach. The value-based approach defines privacy as a human right as part of society’s moral value system. Following this approach, privacy can be conceptualized as a right or a commodity. The cognate-based approach views privacy as related to the individual’s mind, perceptions, and cognition.

Following this approach, privacy can be conceptualized as state or control.

Among these different perspectives, one relevant perspective for this study is to conceptualize privacy as control. Following this perspective, privacy can be viewed as “a process of boundary regulation, controlling how much (or how little) contact an individual maintains with others” (Derlega and Chaikin, 1977, p. 102). In other words, privacy as control represents the interpersonal boundary process through which individuals regulate their interactions with others (Altman, 1975). Therefore, privacy as control is quite relevant to understanding how individuals make their self-disclosure decisions on SNSs while interacting with others. Here, self-disclosure refers to “any message about the self that a person communicates to another” (Wheeless and Grotz, 1976, p. 47). Self-disclosure can support individuals’ boundary regulation by changing the information exchanged (Derlega and Chaikin, 1977).

To understand how individuals regulate their boundaries, we draw upon the CPM theory

(Petronio, 2002). The CPM theory is developed to understand how individuals make self- disclosure decisions within interpersonal relationships. In this theory, the term boundary openingACCEPTED refers to revealing information and theMANUSCRIPT term boundary closure refers to withholding information (Petronio, 2002). The CPM theory proposes three rule elements to manage individuals’ boundaries: boundary rule formation, boundary coordination, and turbulence. In the following paragraphs, we discuss these three elements in the context of SNSs.

Boundary Rule Formation

The CPM theory argues that individuals make self-disclosure decisions on the basis of five criteria (Petronio, 2002): (1) cost–benefit ratio, (2) context, (3) motivations, (4) gender, and

(5) culture. Our study examines all five criteria.

First, the CPM theory argues that individuals use a cost–benefit assessment to determine whether to disclose personal information (Petronio, 2002). Xu et al. (2011) suggest privacy control and privacy risk as two important variables to weigh when individuals balance the costs and benefits involved in self-disclosure. However, Xu et al. (2011) focus on individuals’ privacy perceptions in the context of general websites and do not consider the social aspect of SNSs. Following Jiang et al. (2013), we propose an additional benefit factor, social rewards, referring to the intangible benefits derived from interacting with others. We argue that when individuals decide whether to open their boundaries and disclose personal information, they are likely to evaluate the associated privacy risk, privacy control, and social rewards2. When the outcome is acceptable, individuals are likely to open their boundaries and disclose their information; otherwise, their boundaries remain closed and no information will be disclosed.

Second, the CPM theory suggests that the context influences how privacy rules are established and changed (Petronio, 2002). In other words, specific contexts can have different implications for individuals’ privacy. Therefore, our study focuses on individuals’ privacy concerns in a specific context (i.e., interacting with other contacts on SNSs) and conceptualizes their cost–benefit assessment accordingly. ACCEPTEDThird, the CPM theory proposes that motivational MANUSCRIPT factors also influence the formation of privacy boundary rules (Petronio, 2002). In other words, individuals have different levels of

2 We admit that other factors may be included into cost–benefit assessment in other privacy contexts.

inherent need to maintain their boundaries and disclose their information. Following Xu et al.

(2011), our study uses disposition to value privacy, which represents individuals’ inherent need to maintain their privacy boundaries and protect their privacy. According to Petronio

(2002), motivational factors deal with individuals’ “goals and need for regulating revelation and concealment.” Therefore, it is appropriate to use disposition to value privacy to represent

SNS users’ motivation in the context of our study3.

Fourth, gender may also have a significant effect on forming privacy boundary rules

(Petronio, 2002). Specifically, men and women may have different perspectives with regard to maintaining their privacy (Petronio, 2002). Our study includes gender as well as other demographic variables as control variables.

Finally, culture also has privacy values that are the foundation for forming boundary rules

(Petronio, 2002). Culture orders the privacy expectations of individuals (Hunter, 1991) through which individuals open or close their boundaries. Our study examines how SNS users from different cultures form their privacy rules, selecting the US and China as the context. We use Hofstede’s (2001) cultural dimensions, and these are discussed in more detail below4.

Boundary Coordination

After information is disclosed on SNSs, it becomes accessible by both data subjects (i.e., those disclosing information) and data recipients (i.e., others on SNSs). In such a context, both data subjects and data recipients are responsible for keeping the information private

(Petronio, 2002). The CPM theory suggests that when making self-disclosure decisions,

ACCEPTED MANUSCRIPT 3 We admit that motivation factors may be represented differently in other privacy contexts.

4 Based on the CPM theory, culture—as one of the five criteria influencing individuals’ self-disclosure—moderates the relationship between other criteria (e.g., factors from cost–benefit assessments) and self-disclosure. As this is not our main research objective, we put the corresponding analysis into Appendix D, and we briefly discuss these results in the Implications for Theory section.

individuals may consider how the boundary will be coordinated, such as who can access the disclosed information and how the information will be treated (Petronio, 2002). Data subjects and data recipients may negotiate a set of privacy access and protection rules, which form collectively held privacy boundaries. Such a process of negotiation and collective control over privacy boundaries by both data subjects and data recipients refers to boundary coordination.

Therefore, boundary coordination is vital before self-disclosure, and individuals may negotiate with others regarding how the information should be accessed and privacy needs to be protected. Such negotiations can articulate how privacy will be collectively maintained by both data subjects and data recipients. In the context of SNSs, we propose that group norms and privacy settings are two boundary coordination mechanisms to ensure privacy after the information is disclosed. Individuals interact with others through platforms of SNSs.

Therefore, our study selects group norms and privacy settings such that group norms focus on how individuals interact with each other and privacy settings determine how individuals interact with SNS platforms.

Through group norms, individuals and their contacts on SNSs have shared values regarding how disclosed information should be treated. In addition, the privacy settings of

SNSs can further ensure that disclosed information will be held in a protective domain.

Therefore, these two factors represent two approaches to control disclosed information and coordinate privacy boundaries. Group norms can coordinate privacy boundaries by ensuring that individuals understand privacy and treat disclosed information in a manner consistent withACCEPTED that of others. Privacy settings help coordinate MANUSCRIPT privacy boundaries by technically dividing SNSs into separate protective domains for interacting with different social circles.

Boundary Turbulence

In their social interactions offline, individuals interact with their social circles (e.g., relatives, colleagues, and schoolmates) and have different roles in those social circles. A role is defined as the perception individuals have about themselves when they interact with a certain group in a specific environment (Hall, 1971). As social circles have inconsistent role expectations, individuals need to switch to corresponding roles when interacting with certain social circles. To manage different roles, individuals create and maintain boundaries for their social circles to regulate their social activities (Nippert-Eng, 1996). Individuals then decide what information it is appropriate to disclose on the basis of the social circle concerned.

However, on SNSs, individuals may interact with all of their social circles. In such a context, individuals’ social interactions are not constrained by temporal and spatial separations and the boundaries of their different social circles become blurred. To deal with inconsistent role expectations, individuals may need to switch between roles in different social activities on SNSs (Zhang et al., 2011). As the number of roles to handle increases, the boundary coordination mechanisms may not work. In such a context, a boundary may become turbulent and privacy can be violated (Petronio, 2002). In the context of SNSs, we propose that role conflict and role overload can contribute to boundary turbulence. These two factors are selected such that role overload focuses on the quantities of roles that individuals need to deal with on SNSs, whereas role conflict deals with how these roles are different qualitatively. Role overload and role conflict thus focus on the quantitative and qualitative aspects of roles on SNSs, respectively.

Role overload is defined as the degree to which individuals lack sufficient resources to dealACCEPTED with their different role expectations (Bostrom, MANUSCRIPT 1981). Individuals have limited cognitive resources, which constrain the amount of information that can be processed within a certain period of time (Carpenter et al., 1994). Within social circles, there are certain expectations of individuals’ role, and these role expectations influence how individuals form their privacy

boundaries when making self-disclosure decisions. Therefore, individuals with quite many roles may feel that they do not have adequate resources to deal with the expectations associated with those roles (Kahn et al., 1964). As a result, their privacy boundaries may become turbulent and difficult to coordinate.

Role conflict is defined as the degree to which individuals’ different role expectations

(e.g., norms, standards, or values) are incompatible and incongruent (Guimaraes and Igbaria,

1992). Individuals are evaluated by their social circles on the basis of corresponding role expectations. As the number of social circles increases, it can be challenging to fulfill those inconsistent role expectations simultaneously. For example, individuals not only may be evaluated by how much time they spend on fishing by a fishing group (i.e., an interest group) but may also be judged on the basis of how many lines of code they program by a programming group (i.e., a professional group). Individuals may find it challenging to excel in all the activities inherent in their various role expectations. Inconsistent role expectations can thus make privacy boundaries blurred and turbulent.

To summarize, our study applies Petronio’s (2002) CPM theory as a framework to explain how SNS users’ privacy boundary rules are influenced by boundary coordination and turbulence and how their privacy boundary rules influence their self-disclosure decisions. To the best of our knowledge, Petronio’s CPM theory is the only framework covering this whole process. Other theories, while helpful to understand privacy, cannot fully explain this whole process. For example, the privacy calculus perspective (Dinev and Hart, 2006) can be helpful to explain how SNS users conduct cost–benefit assessment associated with self-disclosure, butACCEPTED it does not examine how individuals’ cost –MANUSCRIPTbenefit assessment is influenced by their privacy boundary coordination or turbulence on SNSs.

Hofstede’s National Cultural Dimensions

National culture refers to “the collective programming of the mind that distinguishes the members of one group or category of people from another” (Hofstede, 2001, p. 9). Our study focuses on two of Hofstede’s (2001) cultural dimensions: individualism/collectivism and uncertainty avoidance. The masculinity/femininity dimension scores are relatively close between the US and China. Therefore, any difference between the US and China in our analysis cannot be attributed to this dimension. We also do not include the long-term orientation dimension, which has received criticism in the previous literature (Feng, 2003;

Redpath and Nielsen, 1997). Our approach of excluding masculinity/femininity and long- term orientation is also consistent with that of Chen and Zahedi (2016), who compare individuals’ Internet security perception between the US and China. Power distance can be helpful to understand individuals’ privacy concerns but is less relevant in understanding the effect of social interactions in the context of this study. Therefore, these three others of

Hofstede’s dimensions are not examined in our study.

Uncertainty avoidance refers to the extent to which individuals feel threatened by ambiguous situations (Hofstede, 2001). Individuals with high uncertainty avoidance avoid deviant behaviors and all types of risks. When individuals disclose their information on

SNSs, information flows out of their boundaries and become accessible by others. In such a context, disclosed information is subject to potential risks (e.g., misuse by others). Therefore, individuals with high uncertainty avoidance may choose not to disclose their information.

Individualism/collectivism refers to whether individuals prioritize their interests over those of a group (Hofstede, 2001). Individuals with high individualism tend to focus on “doing theirACCEPTED own thing” and de-emphasize the interests MANUSCRIPT of the group. On the other hand, individuals with high collectivism are interdependent and emotionally attached to groups. In the context of our study, those with high collectivism may rely more on group norms to coordinate their privacy boundaries because they are more closely connected with others on SNSs.

In addition to Hofstede’s (2001), there are other models of national culture (see Table 2 from Myers and Tan (2002) for a summary). For example, Schwartz (1994) proposes two culture-level dimensions. The first dimension is autonomy/conservatism, and the second dimension is hierarchy and mastery/egalitarian commitment and harmony with nature. For the first dimension, the autonomy culture favors individual thought, feeling, and action, whereas the conservatism culture favors propriety and harmony in social and person-to-group relationships. The second dimension deals with whether a culture emphasizes mastering the surrounding environment or focuses on others’ welfare and harmony with nature.

Hall and Hall (1987) classify culture into high context versus low context. In countries with high-context culture (e.g., Japan), individuals share information continuously and are sensitive to body language. Information communicated, how it is communicated, and information not communicated are all important. On the other hand, in low-context cultures such as Western cultures, information is communicated with rules, procedures, and formal channels such that it can be understood.

Another example is from Trompenaars (1993), who proposes that there are five orientations explaining individuals’ relationship with each other: universalism/particularism, individualism/collectivism, neural/emotional, specific/diffuse, and achievement/ascription.

For universalism/particularism, individuals with the universalist approach believe that they should always follow the one good way, whereas those with the particularist reasoning feel that friendship may come first because it has special obligations. The concept of individualism/collectivism is consistent with that of Hofstede (2001). The neutral/emotional orientationACCEPTED deals with whether individuals feel MANUSCRIPT that social interactions should be objective, or can involve expressing emotion. For the specific/diffuse orientation, a specific relationship is prescribed by a contact, whereas a diffuse relationship involves a real and personal contact.

Finally, in an achievement culture, individuals are judged on the basis of their

accomplishments, whereas in an ascription culture, individuals’ status is attributed by factors such as birth, kinship, and connections.

Although these different models of culture can all be helpful to understand individuals’ privacy decisions in SNSs, our study chooses Hofstede (2001) for three main reasons. First, culture is complex and multidimensional in nature (Myers and Tan, 2002). Therefore, cultural models proposing a single dimension such as Hall and Hall (1987) cannot provide sufficient understanding of culture.

Second, our study aims to examine individuals’ self-disclosure decision-making on SNSs.

As discussed above, self-disclosure is subject to potential risks and its consequences are uncertain. Therefore, uncertainty avoidance, which deals with how individuals feel threatened by ambiguous situations, can be quite helpful to understand how individuals perceive risks associated with self-disclosure. However, few other cultural models assess how individuals perceive an ambiguous environment and respond to its associated risks.

Third, Hofstede’s (2001) cultural model is widely used in the previous privacy literature

(see Table A.2 in Appendix A). By applying Hofstede’s cultural model, our study can advance our understanding of the role of these cultural dimensions in individuals’ privacy perception and self-disclosure decision-making.

Hofstede’s (2001) cultural model is not without limitations. First, Hofstede assumes that cultural differences are aligned with the boundaries of nations. This assumption can be problematic (Jones and Alony, 2007; Myers and Tan, 2002): cultural groups can exist across many nations, and there can also be cultural differences within nations (Harris and Davison, 1999;ACCEPTED Peppas, 2001). Second, with globalization, MANUSCRIPT individuals have more opportunities to interact with others from different cultural backgrounds (Groeschl and Doherty, 2000). The use of new technology such as social media and mobile devices has resulted in cultural convergence (Jenkins, 2006). Therefore, it may be problematic to assume individuals’

cultures simply based on the nations they come from. To address these limitations, we conduct analysis and make comparisons depending on not only participants’ nations but also their espoused cultural dimensions (see Data Analysis and Results in the Methodology section).

To summarize, previous literature has not well examined the effect of individuals’ social circles on their self-disclosure on SNSs. Further, more studies are needed to understand how relationships are different across espoused cultural dimensions. To fill these gaps in the literature, our study examines how SNS users form their privacy boundaries and make their information disclosure decisions when interacting with different social circles on SNSs based upon the CPM theory, as well as the role of culture in individuals’ privacy boundary formation drawn upon Hofstede’s (2001) cultural dimensions.

Hypotheses Development

In this section, we develop our hypotheses on the basis of the CPM theory and the previous literature on privacy. Figure 1 depicts the research model. Here, culture moderates the relationship between boundary coordination/turbulence and cost–benefit assessment. In other words, culture has a moderating effect on how individuals form their privacy boundary rules. In the sections below, we define the constructs in our model and describe each hypothesis in more detail.

Privacy Boundary Rule and Information Disclosure

Cost–Benefit Assessment

Our study proposes perceived privacy risk, perceived privacy control, and social rewards as importantACCEPTED factors in cost–benefit assessment. MANUSCRIPT Social rewards are defined as the pleasure, satisfaction, and gratification derived from interacting with others (Eisenberger et al., 1990).

Jiang et al. (2013) propose social rewards as the positive aspect of privacy trade-off and find that social rewards can lead to greater self-disclosure. On SNSs, friendly self-disclosure is 19

necessary to establish and maintain rewarding relationships (Ben-Ze’ev, 2003). When individuals perceive that they can derive social rewards by interacting with others, they are likely to maintain or further develop those relationships. As a result, such individuals are likely to increase their self-disclosure toward the sources of the rewarding relationships

(Lawler and Thye, 1999). To summarize, we propose:

H1: Social rewards are positively related to the intention to self-disclose.

Privacy risk refers to the degree to which individuals expect a high potential loss resulting from revealing their personal information (Dowling and Staelin, 1994). The previous literature has shown that privacy risk is positively related to privacy concern and negatively related to self-disclosure intention (Dinev and Hart, 2005; Krasnova et al., 2010; Malhotra et al., 2004). Before disclosing information on SNSs, individuals may evaluate the associated risks (e.g., personal information misuse) and the severity of those risks. When individuals perceive a high level of privacy risk, individuals are less likely to disclose their information.

To summarize, we propose:

H2: Perceived privacy risk is negatively related to the intention to self-disclose.

Perceived control refers to the degree to which individuals perceive that they can control their personal information (Malhotra et al., 2004). The previous literature has shown that privacy control is negatively related to individuals’ privacy concerns (Dinev and Hart, 2004;

Xu et al., 2011). On SNSs, when individuals feel that they can control their personal information, they probably feel that their disclosed information is less likely to be misused by others, thus lowering the privacy risk associated with self-disclosure. Further, with a high levelACCEPTED of privacy control, individuals feel more MANUSCRIPT secure and are more comfortable about opening their privacy boundaries. In such a context, individuals’ social interactions with others are probably facilitated and they are more likely to derive social rewards. To summarize, we propose:

H3a: Perceived privacy control is positively related to social rewards.

H3b: Perceived privacy control is negatively related to privacy risk.

Disposition to Value Privacy

Disposition to value privacy is defined as “an individual’s general tendency to preserve his/her private information space or to restrain disclosure of personal information across a broad spectrum of situations and contexts” (Xu et al., 2011, p. 805). It is a personality attribute reflecting individuals’ inherent need to manage the opening and closing of their privacy boundaries. Individuals with a higher disposition to value privacy tend to care more about their personal boundaries and need more control over their own information (Xu et al.,

2011). On SNSs, such individuals are less likely to disclose their personal information. To summarize, we propose:

H4a: Disposition to value privacy is negatively related to the intention to self-disclose.

Disposition to value privacy also influences individuals’ assessment of privacy control and privacy risk (Xu et al., 2011). With a higher disposition to value privacy, individuals are more likely to feel concern regarding their privacy boundaries. These individuals usually desire more control over their disclosed information. Therefore, they are more likely to perceive that the control they have over their disclosed information is not sufficient, thus leading to a lower level of perceived privacy control on SNSs. Further, with the same level of privacy protection and control on SNSs, individuals with a higher disposition to value privacy probably feel a higher level of risk associated with their disclosed information. Therefore, we hypothesize: ACCEPTEDH4b: Disposition to value privacy is negatively MANUSCRIPT related to privacy control. H4c: Disposition to value privacy is positively related to privacy risk.

Boundary Coordination and Cost–Benefit Assessment

Our study proposes group norms and perceived effectiveness of privacy settings as boundary coordination mechanisms on SNSs. Group norms refer to the degree of shared values or goals that individuals perceive between themselves and others on SNSs (Bagozzi and Dholakia, 2002). To disclose information, individuals need to form their intention to actively participate on SNSs. To form such an intention, individuals need to share common goals or values with others on SNSs (Bagozzi and Dholakia, 2002). The previous literature has found that group norms are positively related to group intentions (Bagozzi and Dholakia,

2002).

Group norms represent the prototypical properties of a group and inform individuals about what activities are expected and appropriate within the group (Turner, 1982). Group norms can have different effects based on specific values or goals shared among individuals. In this study, we use group norms to reflect a specific goal related to privacy protection: to ensure that individuals’ disclosed information on SNSs is not released. In other words, we use group norms to refer to the norms of privacy for every user to follow on SNSs, not the norms from different social circles. On SNSs, when individuals perceive that others do not release disclosed information, they probably infer a norm regarding how disclosed information should be treated. In such a context, individuals and others have well-defined group norms regarding how disclosed information should be treated. As a result, individuals are more likely to perceive that their disclosed information is under control. Further, individuals may be attracted to certain groups by social rewards (Hackman, 1992). When individuals are attracted to groups to obtain social rewards, they are more likely to observe others’ behaviors andACCEPTED follow the appropriate behaviors (i.e., group MANUSCRIPT norms). In other words, individuals may follow certain group norms to fulfill their needs of social rewards (Hackman, 1992). In the context of SNSs, individuals may participate in SNSs to obtain social rewards. By following

group norms regarding how the disclosed information should be treated, they are more likely to derive satisfaction and pleasure through interacting with others.

Individuals with high collectivism (i.e., Chinese) usually emphasize “attending to and fitting in with others and the importance of harmonious interdependence and relationships with them” (Markus and Kitayama, 1991, p. 224). In other words, they put more emphasis on groups. On SNSs, when individuals with high collectivism perceive that protecting others’ privacy is a shared norm and that everyone else follows this norm, they are more likely to follow this norm of privacy protection to maintain social relationships with others. In such a context, the shared norm of privacy protection probably helps individuals with high collectivism feel greater control over their privacy and helps them derive more social rewards through interacting with others. Therefore, we hypothesize that:

H5a: Group norms are positively related to social rewards, and this relationship is stronger in cultures with high collectivism (US < China).

H5b: Group norms are positively related to privacy control, and this relationship is stronger in cultures with high collectivism (US < China).

Privacy settings can help individuals control their information on SNSs. For example, on

Facebook, individuals can decide who can access their disclosed information (e.g., the public, all their friends, or a customized group of friends). In such a context, the privacy settings on

Facebook allow individuals to specify boundary opening or closing for a certain group.

Therefore, we define perceived effectiveness of privacy settings as the degree to which individuals feel that the privacy settings on SNSs are capable of protecting their privacy. Xu et al.ACCEPTED (2011) propose a similar concept called theMANUSCRIPT perceived effectiveness of privacy policy to describe the extent to which consumers feel that a posted privacy policy can provide accurate and relevant information regarding an organization’s privacy practices. They find that the perceived effectiveness of a privacy policy is positively related to privacy control.

SNSs’ privacy settings, when effective, can ensure that disclosed information will only be accessible in a protective domain. As a result, individuals probably feel that their disclosed information is better controlled. Further, limiting information disclosure within a certain domain can facilitate individuals’ social interactions with others. For example, individuals may make fishing pictures only accessible to their friends who also love fishing. Through interacting with those friends, they are more likely to derive pleasure.

Individuals with high individualism tend to focus on “doing their own thing” and place greater emphasis on their own interests. Because individuals decide for themselves as to how they want to set up their privacy settings on SNSs, those with high individualism (i.e., low collectivism) may rely more on privacy settings to protect their privacy and facilitate social interactions with others. In such a context, more effective privacy settings on SNSs can help individuals with high individualism better control their privacy and derive more social rewards. Therefore, we propose that:

H6a: Perceived effectiveness of privacy settings is positively related to social rewards, and this relationship is stronger in cultures with low collectivism (US > China).

H6b: Perceived effectiveness of privacy settings is positively related to privacy control, and this relationship is stronger in cultures with low collectivism (US > China).

Boundary Turbulence and Cost–Benefit Assessment

On SNSs, individuals may interact with contacts from different social circles. Because these social circles often have inconsistent role expectations, individuals may feel overwhelmed managing those conflicting roles. Our study proposes role conflict and role overloadACCEPTED as mechanisms of boundary turbulence. MANUSCRIPT Individuals interact with their various social circles such as friends, classmates, and relatives on SNSs. As individuals’ social circles become more diverse, they may face inconsistent role expectations from their various social circles (i.e., role conflict) and find managing these different expectations challenging (i.e.,

role overload). As a result, their privacy boundaries may become turbulent and difficult to coordinate. In such a context, individuals are likely to perceive that disclosing personal information may lead to unfavorable outcomes and thus are likely to feel a high level of privacy risk.

Further, individuals with high uncertainty avoidance (i.e., Americans) feel less comfortable in an ambiguous environment. Therefore, when these individuals perceive that they need to deal with inconsistent expectations from their various social circles (i.e., role conflict) and find it challenging to meet these different expectations (i.e., role overload), they may feel that the social environment on SNSs is ambiguous because they are not sure who will view their disclosed information and how their disclosed information will be used. In such a context, they are more likely to perceive a more severe risk to their privacy. On the other hand, individuals with low uncertainty avoidance (i.e., Chinese) are more tolerant of risk. As a result, they may feel less concerned about their privacy when experiencing role conflict and role overload. Therefore, we propose:

H7: Role conflict is positively related to privacy risk, and this relationship is stronger in cultures with high uncertainty avoidance (US > China).

H8: Role overload is positively related to privacy risk, and this relationship is stronger in cultures with high uncertainty avoidance (US > China).

Control Variables

Following on the previous literature, we control for several variables that potentially influence intention to self-disclose. First, we control for gender (Sheehan, 1999), age (Culnan,ACCEPTED 1995), and previous privacy experiences MANUSCRIPT (Smith et al., 1996). Second, we control for the number of contact types (because individuals will have more roles with more types of contacts). Finally, we control for trust toward other SNS users and trust toward SNSs, because SNS users’ self-disclosure decisions can be influenced by the trust that individuals

have with not only the SNS but also with their friends on the SNS. Here, trust toward SNSs may increase individuals’ perceived safety within the online environment of SNSs (Gerlach et al., 2015).

Methodology

Data Collection Procedures and Participants

Survey companies were employed to recruit SNS users from the US and China. Survey companies maintain various channels to recruit a variety of samples. Several approaches such as quality assurance questions were used to detect fraudulent or duplicate answers. In total, we received 381 valid responses from the US and 450 valid responses from China5. The demographic data of participants are shown in Table 1, and their activities on their most frequently used SNSs are shown in Table 2. We also show cultural indices for the US and

China from Hofstede’s (2001) cultural model (Table 3).

Table 1. Sample Demographic Information US (N = 381) China (N = 450) Gender 69.82% Female 53.78% Female Age 19 or below 7.61% 26.22% 20-29 36.22% 32.89% 30-39 15.22% 23.33% 40-49 14.44% 12.22% 50 or above 26.51% 5.33% Education Primary school 1.57% 1.78% High school 18.37% 38.23% Associate degree 43.04% 24.00% Bachelor degree or above 37.02% 36.00%

Table 2. Activities on Most Frequently Used SNSs US (N = 381) China (N = 450) Most Frequently Used SNSs Facebook (80.58%) WeChat (75.11 %) Twitter (7.09%) Weibo (12.22%) ACCEPTED InstagramMANUSCRIPT (6.04%) QQ (11.56%)

5 Two survey companies were employed to collect data from the US and China, respectively. For the US data collection, 825 participants were invited, 597 took the survey (response rate: 72.36%), and 381 responses were valid (63.82%). For the China data collection, 1000 participants were invited, 692 took the survey (response rate: 69.20%), and 450 responses were valid (valid rate: 65.03%).

Pinterest (1.84%) Renren (0.89%) Other (4.45%) Other (0.22%) Usage Experience Less than 1 month 1.57% 2.22% 1-6 months 1.57% 3.56% 6-12 months 3.41% 8.44% 1-2 years 8.14% 23.33% More than 2 years 85.30% 62.44% Types of Contacts Relatives 84.00% 80.22% Supervisors 12.86% 28.89% Colleagues 45.93% 62.44% Subordinates 9.71% 18.89% Schoolmates 73.25% 88.00% Friends 90.81% 92.00% Others 22.05% 5.33% Number of Contacts Below 50 16.01% 15.56% 51-100 14.70% 25.33% 101-150 13.12% 20.00% 151-200 14.96% 11.78% More than 200 41.21% 27.33%

Table 3: Cultural Dimensions Uncertainty Avoidance Individualism US 46 91 China 30 20 Source: https://geert-hofstede.com/united-states.html

Measurement

Our items were adapted from the previous literature (See Appendix B). Specifically, role conflict was adapted from Zhang et al. (2011), Rutner et al. (2008), and Rizzo et al. (1970); role overload was adapted from Zhang et al. (2011); perceived effectiveness of privacy settings, perceived control, perceived risk, disposition to value privacy, and previous privacy experiences were adapted from Xu et al. (2011); intention to self-disclose was adapted from

Malhotra et al. (2004); social rewards were adapted from Jiang et al. (2013); trust toward the

SNSACCEPTED and trust toward other SNS users were adaptedMANUSCRIPT from Krasnova et al. (2010); and power distance, uncertainty avoidance, and collectivism were adapted from Chen and Zahedi

(2016). All items were measured with a 7-point Likert scale ranging from Strongly Disagree to Strongly Agree. The number of contact types was measured using seven items, each of 27

which evaluated whether participants had a certain type of contact (Table 2) and the participants answered with either Yes or No. Then, a sum score was calculated.

Following Bagozzi and Dholakia (2002), group norms were measured with two 7-point items indicating the degree of shared goals between the individuals and their close contacts on SNSs. The first item measured the strength of shared goals for close contacts (referred to as “GNC”), whereas the second item measured the strength of shared goals for individuals themselves (referred to as “GNS”). These two items were measured with following procedures. First, participants were asked to list five contacts that they interact most frequently on SNSs. Second, the following statement was introduced: “On social networking sites, ‘Ensure that users’ information should not be released’ can be considered a shared goal among users. For five of your closest contacts listed above, please estimate the strength with which each holds that goal.” Participants thus rated each of their contacts with a 7-point

Likert scale ranging from very weak to very strong. The average score of ratings for five contacts was used as GNC for each participant. Finally, participants rated the degree to which they hold the shard goal themselves, and the rating was used as GNS.

Data Analysis and Results

Because all the variables were collected in one survey, we first assessed the potential threat of common method bias (CMB) with a pooled sample (Podsakoff et al., 2003). First, a

Harmon one-factor analysis was conducted. The results showed that four factors were present and that the most variance explained by one factor was 24.28%. Second, following Podsakoff et al. (2003), a common method factor including all the principal factors’ indicators was includedACCEPTED in the partial least squares (PLS), and MANUSCRIPT we calculated the variances of indicators explained by the principal factor and by the method. Our results showed that the average substantively explained variance of the indicator was .74, whereas the average method–based variance was .004. The ratio of substantive variance to method variance was around 284:1. In

addition, most method factor loadings were not significant. Therefore, we concluded that

CMB was unlikely to be a serious concern for our study.

Our model was tested with PLS. The bootstrap resampling method (using 1,000 samples) was applied to determine the significance of the paths. PLS was used for several reasons.

First, PLS is appropriate for highly complex predictive models (Chin, 1998). Previous literature applying PLS (e.g., Kim and Benbasat, 2006) has found that PLS can test complex relationships while avoiding inadmissible solutions and factor indeterminacy. Therefore, PLS was more appropriate to accommodate a large number of constructs and relationships in our model. Second, Shapiro–Wilk tests were significant, showing that the measurements were not normally distributed. According to Hair et al. (2014), PLS is more appropriate with non- normally distributed data. Third, PLS is suitable for exploratory research (Gefen et al., 2011).

Because our study develops a new model to examine individuals’ privacy in decision-making in SNSs, our study is exploratory in nature rather than testing an established theory.

Therefore, PLS is more appropriate for our study.

We first evaluated the measurement model. As shown in Table C.1 in Appendix C, each item loaded significantly on its respective construct and none of the loadings were below .50

(Hulland, 1999). In addition, Cronbach’s alpha and composite reliabilities (CRs) were more than 0.70, and the average variance extracted (AVE) was more than .50 (Table 4). Therefore, convergent validity was supported. Discriminant validity was also confirmed by ensuring that the correlations between constructs were less than .85 (Brown, 2006), and for each construct, the square root of its AVE exceeded all correlations between that factor and any other constructACCEPTED (Table 5). Therefore, our measures demonstratedMANUSCRIPT good psychometric properties. Table 4. Cronbach’s Alpha, CR, and AVE US China Construct Cronbach’s Cronbach’s CR AVE CR AVE Alpha Alpha Disposition to Value Privacy .81 .89 .72 .79 .88 .71

Group Norms .79 .91 .83 .84 .92 .86 Privacy Control .90 .93 .77 .76 .85 .58 Perceived Effectiveness .89 .93 .82 .81 .89 .73 Privacy Risk .90 .93 .76 .89 .92 .74 Role Conflict .86 .91 .71 .75 .84 .57 Role Overload .96 .97 .83 .87 .90 .61 Social Rewards .87 .92 .79 .80 .88 .72 Intention to Self-disclose .95 .96 .87 .89 .93 .76 Power Distance .80 .88 .72 .72 .76 .54 Uncertainty Avoidance .86 .90 .63 .85 .90 .63 Collectivism .87 .90 .65 .88 .91 .67 Previous Privacy Experiences .79 .86 .69 .68 .82 .61 Trust Toward SNSs .94 .95 .77 .92 .94 .72 Trust Toward Other Users .94 .95 .76 .92 .94 .71

Table 5. Correlation Between Constructs and Square Root of AVEs (on Diagonal) US 1 2 3 4 5 6 7 8 9 10 11 12 1 Disposition to Value .85 Privacy 2 Group Norms .25 .91 3 Privacy Control .15 .28 .88 4 Perceived .16 .27 .65 .91 Effectiveness 5 Privacy Risk .29 - - .02 .87 .10 .05 6 Privacy Control .13 - .19 .16 .46 .84 .01 7 Role Overload .16 - .26 .25 .54 .70 .91 .02 8 Social Rewards .12 .37 .49 .51 .07 .32 .33 .89 9 Intention to Self- - .19 .35 .34 .09 .34 .34 .62 .93 disclose .03 10 Previous Privacy .22 - - - .41 .47 .51 .12 .17 .83 Experiences .02 .04 .05 11 Trust Toward SNSs .17 .50 .48 .59 - .12 .19 .56 .43 -.05 .88 .04 12 Trust Toward Other .21 .44 .45 .43 .06 0.20 .24 .49 .41 .02 .63 .87 Users China 1 2 3 4 5 6 7 8 9 10 11 12 1 Disposition to Value .84 Privacy 2 Group Norms .21 .93 3 ACCEPTEDPrivacy Control - .36 .76 MANUSCRIPT .01 4 Perceived .10 .35 .47 .85 Effectiveness 5 Privacy Risk .31 .02 - - .86 .11 .05

6 Privacy Control .03 .03 - .02 .25 .75 .03 7 Role Overload .06 - - - .30 .51 .78 .01 .08 .03 8 Social Rewards .13 .39 .41 .41 - .04 .03 .85 .13 9 Intention to Self- - .25 .26 .36 - .01 - .44 .87 disclose .07 .31 .02 10 Previous Privacy .14 .00 - - .37 .22 .26 - -.08 .78 Experiences .16 .06 .06 11 Trust Toward SNSs .20 .50 .46 .48 - .09 .05 .49 .32 .00 .85 .03 12 Trust Toward Other .06 .48 .46 .43 - -.01 - .51 .32 -.06 .67 .84 Users .14 .03

We then tested the structural paths in our model (Table 6). For both samples, most of the hypotheses (i.e., H1, H3a, H3c, H4a, H5a, H6a, H6b, H7, and H8) were supported, thus strongly supporting our model. In addition, for the US sample, privacy control was negatively related to privacy risk (ß = -23, p < .001). For the China sample, privacy risk was negatively related to intention to self-disclose (ß = -25, p < .001), disposition to value privacy was negatively related to perceived control (ß = -10, p < .05), and group norms were positively related to privacy control (ß = 24, p < .001).

Table 6. Structural Model Hypothesis US China T- Diff. Supported? value Sig. H1: Social rewards → Self-disclosure .47*** .33*** H2: Privacy risk → Self-disclosure - .05 .25*** H3a: Privacy control → Social rewards .23*** .21*** H3b: Privacy control → Privacy risk -.23*** -.09 H4a: Disposition to value privacy → -.17** -.10* Self-disclosure H4b: Disposition to value privacy → .03 -.10* Privacy control H4c: Disposition to value privacy → .23*** .29*** PrivacyACCEPTED risk MANUSCRIPT H5a: Group norms → Social rewards .23*** .23*** -0.19 ns No H5b: Group norms → Privacy control .10 .24*** -8.80 sd Yes H6a: Perceived effectiveness → Social .31*** .23*** 5.11 >*** Yes rewards H6b: Perceived effectiveness → .62*** .40*** 14.64 >*** Yes Privacy control 31

H7: Role conflict → Privacy risk .16** .13* 1.78 ns No H8: Role overload → Privacy risk .45*** .22*** 14.14 >*** Yes Note. * p < .05, ** p < .01, *** p < .001; ns = not significant; sd = structurally different (one path is significant and the other is not); Diff. Sig. = Different Significantly; Supported? = Is the hypothesis supported?

We listed R2 in our structural model. Based on the results in Table 7, the proposed relationships dealing with boundary coordination/turbulence can explain a large amount of variance in privacy risk and social rewards. Further, our proposed variables (i.e., social rewards, privacy risk, and disposition to value privacy) can explain a large amount of unique variance of self-disclosure in addition to control variables. Finally, the values of R2 in the US sample are higher than those in the China sample, thus indicating that our model can better explain Americans’ self-disclosure decisions.

Table 7. R2 in Structural Model R2 Dependent Variable US China Self-disclosure (the Control-variable-only Model) 30.3% 14.2% Self-disclosure (the Full Model) 46.0% 29.4% Privacy control 43.3% 27.4% Privacy risk 39.5% 19.8% Social Rewards 35.3% 27.3%

We then examined the moderating effects of culture by comparing the path coefficients between the US and China samples with the formula of Keil et al. (2000):

Path coefficientGroup1 − Path coefficentGroup2 t = (m − 1)2 (n − 1)2 1 1 [√ × SE2 + × SE2 ] × [√ + ] (m + n − 2) Group1 (m + n − 2) Group2 m n where m is the sample size of Group 1 and n is the sample size of Group 2.

The cultural moderation in H5b was supported, thus indicating that group norms had a stronger effect on privacy control for the Chinese (high collectivism). The cultural moderationACCEPTED in H6a and H6b was also supported, MANUSCRIPT thus indicating that perceived effectiveness of privacy settings had a stronger effect on social rewards and privacy control for the

Americans (low collectivism). Finally, the cultural moderation in H8 was also supported, thus indicating that role overload had a stronger effect on privacy risk for the Americans (high

uncertainty avoidance). These findings are presented in Table 6.

Post Hoc Analysis: The Espoused Culture’s Moderating Effects in the Pooled Sample

To address the limitations of Hofstede’s (2001) cultural model, we then examined the espoused culture’s moderating effects in our pooled sample following Chen and Zahedi

(2016). Specifically, the pooled sample was divided into low versus high uncertainty avoidance and collectivism. The results are shown in Table 8.

Table 8. Post Hoc Analysis of the Moderation of Espoused Cultural Dimensions Hypothesis Uncertainty Avoidance Collectivism T- Diff. T- Diff. Low High Low High value Sig. value Sig. H5a: Group norms → .05 .19** -8.20 sd .09 .18** -5.29 sd Social rewards H5b: Group norms → .15** .15** 0 ns .13** .16** -2.06 <* Privacy control H6a: Perceived .34** .21** effectiveness → Social 8.27 >*** .28*** .29*** -0.43 ns * * rewards H6b: Perceived .55** .52** effectiveness → Privacy 2.04 >* .60*** .46*** 9.15 >*** * * control H7: Role conflict → .22** .15** -4.25 <*** .15** .22*** -3.99 <*** Privacy risk * H8: Role overload → .34** .33** 0.62 ns .38*** .29*** 5.51 >*** Privacy risk * * Note. * p < .05, ** p < .01, *** p < .001, ns = not significant, sd = structurally different (one path is significant and the other is not), Diff. Sig. = Different Significantly

The results showed that espoused uncertainty avoidance moderated the path from role conflict and privacy risk (H7). High uncertainty avoidance had a higher path coefficient for role conflict → privacy risk. Espoused collectivism had a moderating effect on group norms

→ social rewards (H5a), group norms → privacy control (H5b), and perceived effectiveness

→ privacy control (H6b). Here, high collectivism had a higher coefficient for group norms → socialACCEPTED rewards and group norms → privacy control, MANUSCRIPT whereas low collectivism had a higher coefficient for perceived effectiveness → privacy control.

Discussion

Our study examines how individuals make their information disclosure decisions as well as how individuals’ boundary rule formation is influenced by boundary coordination and turbulence in the context of SNSs. Our model is tested with American and Chinese SNS users, and the results strongly support our model. Specifically, our results show that social rewards are positively related to the intention to self-disclose, whereas privacy risk and disposition to value privacy are negatively related to the intention to self-disclose.

In our comparison between the US and China, the moderating effect in H5b was supported, thus indicating that group norms had a stronger effect on privacy control for the

Chinese than for the Americans. Our post hoc analysis also shows that group norms have a stronger effect on social rewards and privacy control under high collectivism, consistent with

H5a and H5b.

The moderating effect in H6a and H6b was also supported, thus indicating that perceived effectiveness of privacy settings had a stronger effect on social rewards and privacy control for the Americans than for the Chinese. Our post hoc analysis also shows that perceived effectiveness has a stronger effect on privacy control under low collectivism, consistent with

H6b.

The moderating effect in H7 was not supported, thus indicating that the effect of role conflict on privacy risk is consistent between the US and China. However, our post hoc analysis shows that role conflict has a stronger effect on privacy risk under high uncertainty avoidance, consistent with H7. Therefore, future studies are needed to further examine how individuals from different countries form their perceptions of role conflict. ACCEPTEDFinally, the moderating effect in H8 was supported,MANUSCRIPT thus indicating that role overload had a stronger effect on privacy risk for Americans. However, post hoc analysis with high versus low uncertainty avoidance shows that this relationship is not significantly different across the

two subsamples. These results indicate that other contextual factors may be the cause of the differences between the Americans and the Chinese.

Implications for Theory

Our study makes important theoretical contributions. First, our study clarifies the mechanisms of boundary coordination and turbulence in the context of SNSs. We argue that individuals’ privacy boundaries may become blurred and turbulent owing to role conflict

(H7) and role overload (H8) caused by diverse social circles while they can coordinate their boundaries through group norms (H5) and privacy settings (H6). Boundary turbulence and coordination thus jointly determine individuals’ cost–benefit assessment, a part of their privacy boundary rules. These hypotheses are greatly supported for both the American and

Chinese samples and for the subsamples divided by espoused cultural dimensions, thus indicating the generalizability of our model. Therefore, our study contributes to the privacy literature by providing valuable insights regarding how individuals develop their cost–benefit assessment and form their privacy boundaries while interacting with diverse social circles and how SNSs can help protect their users’ privacy and support friendly interactions among users.

By clarifying the mechanisms of boundary coordination and turbulence, our study further extends the previous literature by applying the CPM theory6. For example, Xu et al. (2011) apply the CPM theory to examine individuals’ information privacy concerns toward websites.

Because the context of Xu et al. (2011) is general websites, they do not include social rewards, an important factor of individuals’ cost–benefit assessment in the context of SNSs

(Jiang et al., 2013). Further, their study does not differentiate the mechanisms of boundary coordinationACCEPTED from those of boundary turbulence, MANUSCRIPT and their operationalization of boundary coordination/turbulence focuses on general websites. Our study thus extends that of Xu et al.

6 The CPM theory is a general theoretical framework and its operationalization depends on specific contexts.

by clarifying how boundary coordination and boundary turbulence influence individuals’ cost–benefit assessment in the context of SNSs7. Specifically, our study clarifies how individuals’ diverse social circles result in boundary turbulence.

Second, our study contributes to the previous privacy literature by highlighting the importance of culture (i.e., individualism and uncertainty avoidance dimension) when individuals make their self-disclosure decisions. Culture is an important criterion for self- disclosure decisions (Petronio, 2002) but has not been well examined to understand how relationships are different across espoused cultural dimensions. Specifically, previous literature has mainly treated culture as given or as independent variables, and few studies have measured espoused cultural dimensions and examined how certain relationships are different under various cultural dimensions. Our study shows interesting differences between the US and China as well as across different cultural dimensions. As discussed above, our results provide stronger support for the moderating effect of individualism/uncertainty avoidance on the relationship between boundary coordination/turbulence and cost–benefit assessment. Therefore, our study clarifies how individuals form their privacy boundaries while interacting with diverse social circles under different cultural contexts. We also assess the moderating effect of individualism/uncertainty avoidance on the relationship between boundary rule formation and self-disclosure (refer to Appendix D). However, the results cannot be fully explained by these two dimensions. Overall, the results show that individualism and uncertainty avoidance play an important role when individuals assess the risk and benefits associated with self-disclosure. However, when making self-disclosure decisions,ACCEPTED individuals are not only influenced MANUSCRIPTby individualism and uncertainty avoidance but

7 In other words, our study operationalizes boundary coordination and boundary turbulence focusing on the context of SNSs. Although Xu et al. (2011) also collect data from SNS users, the context of their study is general websites. They do not include any SNS-specific variables or examine the role of social circles.

are also affected by other contextual factors (Chen and Zahedi, 2016). In other words, while it is important to consider culture when examining privacy, culture may be more helpful to explain individuals’ cost–benefit assessment associated with self-disclosure than individuals’ self-disclosure decisions. Our study thus clarifies the role of culture (i.e., individualism and uncertainty avoidance dimension) in the process of self-disclosure.

Implications for Practice

Our study also has important implications for SNS providers and users. For SNS providers, our analyses consistently show that social rewards have a positive effect on intention to self-disclose. Therefore, SNS providers need to support individuals to derive more social rewards through interacting with others. Here, SNS providers can focus on group norms and privacy settings because those two factors enhance social rewards. Regarding group norms, SNS providers can outline of how disclosed information should be treated by others in their policies to facilitate the creation of appropriate norms for interactions. With regard to privacy settings, SNS providers need to provide more effective settings to help individuals coordinate their boundaries. For example, SNS providers should go further than allowing individuals to select who can view the disclosed information, to allow individuals to also customize how others can view comments. Individuals could limit the accessibility of comments so that Contact A can only view Contact B’s comments when they are contacts with each other as well. By limiting interactions within certain social circles, SNSs can thus increase the effectiveness of privacy settings and prevent misuse of disclosed information.

Such settings can also help individuals handle role conflict and role overload. ACCEPTEDSecond, our study shows important cultural MANUSCRIPT differences that provide useful guidelines for SNS providers operating in different cultures. For example, as the resources of SNS providers are limited, they can focus on different factors to support users’ deriving social rewards in different cultural contexts. The results from our post hoc analysis can divide SNS users into

two groups: those from high collectivism culture and those from low collectivism culture. In

China (or high collectivism culture), SNS providers need to facilitate their users to form suitable norms regarding how to treat others’ privacy and disclosed information. In the US

(or low collectivism culture), SNS providers need to offer more effective privacy settings to help their users better deal with their privacy and disclosed information. SNS providers from high uncertainty avoidance cultures also need to be careful about users’ role conflict. For example, they can add new functionalities to help users keep track of social roles on SNSs and allow users to specify how certain messages can be viewed by different social circles.

For SNS users, our study shows that American users may disclose their information to receive social rewards and they do not feel concerned about their privacy risk during self- disclosure. Therefore, American SNS users need to be more mindful of the potential negative outcomes of self-disclosure and understand how to protect their privacy. Based on our results, they can use privacy settings to better control their privacy and support social interactions with others on SNSs.

Limitations and Opportunities for Future Studies

Our study also has several limitations. First, our participants were recruited by survey companies. Although our participants came from diverse backgrounds and most of them had multiple types of contacts on their most frequently used SNSs, our sample may still be biased. Second, to understand the effect of culture, we collected data from the US and China.

We also conducted post hoc analysis by dividing the pooled sample into subsamples according to different cultural dimensions. Nevertheless, our results may still be limited, and futureACCEPTED studies are needed to examine our model MANUSCRIPT with participants from other countries. Our study used Hofstede’s cultural dimensions, and future studies can use cultural dimensions from other theories and perspectives.

We measured group norms following Bagozzi and Dholakia (2002), and only five closest contacts were rated regarding the strength of shared goals. This approach of operationalizing group norms can be limited, and future studies are needed to examine different ways to operationalize group norms.

Our results provide strong support for the relationship between boundary coordination/turbulence and cost–benefit assessment. However, the relationship between boundary rule formation and self-disclosure could not be fully explained by culture (refer to

Appendix D), thus indicating that additional contextual factors may be needed to fully understand privacy across different countries. For example, Chen and Zahedi (2016) argue that philosophical, political, and economic contexts can also influence differences between two countries. Future studies incorporating relevant factors from different contexts are needed to better understand how individuals from different countries perceive privacy.

Recent literature suggests that millennials—young people born after 1980—are more confident, self-expressive, and open to change (Taylor and Keeter, 2010). Future studies could explore whether millennials make self-disclosure decisions differently. Future studies could also examine additional factors that influence boundary coordination and turbulence.

Individual characteristics and relevant contextual factors may also moderate the relationships between boundary coordination/turbulence and cost–benefit assessment. Finally, because privacy boundaries are difficult to coordinate when they become turbulent, future studies are needed to examine how boundary turbulence may influence boundary coordination.

Conclusion

Zilong Liu ([email protected]) is an associate professor in Department of Information Management,ACCEPTED School of Management Science MANUSCRIPT and Engineering, Dongbei University of Finance and Economics, China. He received his Ph.D. in Information Systems from Dongbei University of Finance and Economics. His research interests include information privacy and social media. He has published several studies in academic journals as well as international journals (e.g. Information & Management) and conferences on those topics. Xuequn (Alex) Wang ([email protected]) is a Lecturer in Murdoch University. He received his Ph.D. in Information Systems from Washington State University. His 39

research interests include social media, online communities, knowledge management, and human-computer interaction. His research has appeared in Technological Forecasting & Social Change, Communications of the Association for Information Systems, Journal of Organizational Computing and Electronic Commerce, Behaviour & Information Technology, Journal of Computer Information Systems, and Journal of Knowledge Management. As individuals interact with more types of social circles on SNSs, they may feel that it is challenging to maintain their privacy. This study has developed a research model to examine how individuals make self-disclosure decisions and how SNSs can protect their privacy. The data, collected from the US and China, strongly support our research model regarding the effect of boundary coordination and turbulence. Further, there are interesting cultural differences with regard to individuals’ information disclosure. Future studies are needed to extend our research by incorporating other contextual factors.

ACCEPTED MANUSCRIPT

References Altman, I. (1975). The Environment and Social Behavior: Privacy, Personal Space, Territory, Crowding. Monterey, CA: Brooks/Cole. Bagozzi, R. P, and Dholakia, U. M. (2002). Intentional Social Action in Virtual Communities. Journal of Interactive Marketing, 16 (2), 2-21. Bélanger, F., and Crossler, R.E. (2011). Privacy in the digital age: A review of information privacy research in information systems. MIS Quarterly, 35(4), 1017–1041. Bellman, S., Johnson, E. J., Kobrin, S. J., and Lohse, G. L. (2004). International Differences in Information Privacy Concerns: A Global Survey of Consumers. The Information Society, 20(5), 313-324. Ben-Ze’ev, A. (2003). Privacy, emotional closeness, and openness in cyberspace. Computers in Human Behavior, 19(4), 451–467. Besmer, A., and Lipford, H.R. (2010). Moving beyond untagging: Photo privacy in a tagged world. Proceedings of the SIGCHI Conference on Human Factors in Computing Systems, Atlanta, GA, USA, 1563–1572. Bostrom, R. P. (1981). Role Conflict and Ambiguity: Critical Variables in the User-Designer Relationship. Proceedings of the Seventeenth Annual Computer Personnel Research Conference, 88-112. Boyd, D. M., and Ellison, N. B. (2008). Social network sites: Definition, history, and scholarship. Journal of Computer-Mediated Communication, 13(1), 210–230. Brandtzæg, P. B., Lüders, M., and Skjetne, J. H. (2010). Too Many Facebook “Friends”? Content Sharing and Sociability Versus the Need for Privacy in Social Network Sites. International Journal of Human–Computer Interaction, 26(11-12), 1006-1030. Brown, T. A. (2006). Confirmatory factor analysis for applied research. New York, NY: The Guilford Press. Carpenter, P., Miyake, A., and Just, M. A. (1994). Working Memory Constraints in Comprehension: Evidence from Individual Differences, Aphasia, and Aging, San Diego, CA: Academic Press. Cecere, G., Le Guel, F., and Soulié, N. (2015). Perceived Internet privacy concerns on social networks in Europe. Technological Forecasting & Social Change, 96, 277-287. Chellappa, R. K., and Sin, R. (2005). Personalization versus privacy: An empirical examination of the online consumer's dilemma. Information Technology and Management, 6(2), 181-202. Chen, J. Q., Zhang, R., and Lee, J. (2013). A Cross-Culture Empirical Study of M-commerce Privacy Concerns. Journal of Internet Commerce, 12(4), 348–364. Chen, R. (2013). Living a private life in public social networks: An exploration of member self-disclosure. Decision Support Systems, 55(3), 661-668. Chen, R., and Sharma, S. K. (2015). Learning and self-disclosure behavior on social networking sites: the case of Facebook users. European Journal of Information Systems, 24(1), 93–106. Chen, Y, and Zahedi, F. M. (2016). Individuals' Internet Security Perceptions and Behaviors: Polycontextual Contrasts between the United States and China. MIS Quarterly, 40(1), 205- ACCEPTED222. MANUSCRIPT Chin, W. W. (1998). The partial least squares approach to structural equation modeling. In G. A. Marcoulides (Ed.), Modern methods for business research (pp. 295-336). Mahwah, NJ: Lawrence Erlbaum Associates. Cho, H., Rivera-Sánchez, M., and Lim, S. S. (2009). A multinational study on online privacy: global concerns and local responses. New Media Society, 11(3), 395–416.

Choi, B. C. F., Jiang, Z., Xiao, B., and Kim, S. S. (2015). Embarrassing Exposures in Online Social Networks: An Integrated Perspective of Privacy Invasion and Relationship Bonding. Information Systems Research, 26(4), 675–694. Cullen, R. (2009). Culture, identity and information privacy in the age of digital government. Online Information Review, 33(3), 405-421. Culnan, M. J. (1995). Consumer awareness of name removal procedures: Implication for direct marketing. Journal of Interactive Marketing, 9(2), 10-19. De Wolf, R., Willaert, K., Pierson, J. (2014). Managing privacy boundaries together: Exploring individual and group privacy management strategies in Facebook. Computers in Human Behavior, 35, 444–454. Derlega, V. J., and Chaikin, A. L. (1977). Privacy and self-disclosure in social relationship. Journal of Social Issues, 33(3), 102–115. Dinev, T., Bellotto, M., Hart, P., Colautti, C., Russo, V., and Serra, I. (2005). Internet Users’ Privacy Concerns and Attitudes towards Government Surveillance – An Exploratory Study of Cross-Cultural Differences between Italy and the United States. Proceedings of 18th Bled eConference eIntegration in Action, Bled, Slovenia. Dinev, T., and Hart, P. (2005). Internet privacy concerns and social awareness as determinants of intention to transact. International Journal of Electronic Commerce, 10(2), 7-29. Dinev, T., and Hart, P. (2006). Privacy Concerns and Privacy-Protective Behavior in Synchronous Online Social Interactions. Information Systems Research, 17(1), 61–80. Dowling, G. R., and Staelin, R. (1994). A Model of Perceived Risk and Intended Risk- handling Activity. Journal of Consumer Research, 21(1), 119-134. Eisenberger R., Fasolo P. M., Davis-LaMastro V. (1990). Perceived organizational support and employee diligence, commitment, and innovation. Journal of Applied Psychology, 75(1), 51–59. Feng, T. (2003). A Critique of Hofstede’s Fifth National Culture Dimension. International Journal of Cross Cultural Management, 3(3), 347-368. Garfield, M. J. and Watson, R. T. (1998). The impact of national culture on national information infrastructure. Journal of Strategic Information Systems, 6(4), 313-338. Gerlach, J., Widjaja, T., Buxmann, P. (2015). Handle with care: How online social network providers’ privacy policies impact users’ information sharing behavior. The Journal of Strategic Information Systems, 24(1), 33-43. Guimaraes, T. and Igbaria, M. (1992). “Determinants of Turnover Intentions: Comparing IC and IS Personnel,” Information Systems Research, 3(3), 273-303. Hackman, J. R. (1992). Group influences on individuals in organizations. In M. D. Dunnette & L. M. Hough (Eds.), Handbook of industrial and organizational psychology (2nd ed., Vol. 3, pp. 199 –267). Palo Alto, CA: Consulting Psychologists Press. Hall, D. T. (1971). A Theoretical Model of Career Subidentity Development in Organizational Settings. Organizational Behavior and Human Performance, 6(1), 50-76. Hall, E. T., and Hall, R. M. (1987). Hidden Differences: Doing Business with the Japanese. New York, NY: Doubleday. Hair,ACCEPTED J. F., Hult, G. T. M., Ringle, C. M., and MANUSCRIPTSarstedt, M. (2014). A Primer on Partial Least Squares Structural Equation Modeling (PLS-SEM). Thousand Oaks, CA: Sage. Harris, R. and Davison, R. (1999). Anxiety and involvement: Cultural dimensions of attitudes toward computers in developing countries. Journal of Global Information Management, 7(1), 26-38. Hofstede, G. 2001. Culture’s Consequences: comparing values, behaviors, institutions, and organization across nations. Beverly Hills, CA: Sage. 42

Hulland, J. (1999). User of Partial Least Squares (PLS) in Strategic Management Research: A Review of Four Recent Studies. Strategic Management Journal, 20(2), 195-204. Hunter, J. D. (1991). Culture wars: The struggle to define America. New York, NY: Basic Books. James, T. L., Wallace, L., Warkentin, M., Kim, B. C., and Collignon, S. E. (In Press). Exposing others’ information on online social networks (OSNs): Perceived shared risk, its determinants, and its influence on OSN privacy control use. Information & Management. James, T. L., Warkentin, M., and Collignon, S. E. (2015). A dual privacy decision model for online social networks. Information & Management, 52(8), 893–908. Jenkins, H. (2006). Convergence Culture – Where Old and New Media Collide. New York: New York University Press. Jiang, Z., Heng, C. S., and Choi, B. C. F. (2013). Privacy concerns and privacy-protective behavior in synchronous online social interactions. Information Systems Research, 24(3), 579-595. Jones, M. L., and Alony, A. (2007). The cultural impact of information systems – through the eyes of Hofstede – a critical journal. Issues in Informing Science and Information Technology, 4, 407-419. Kahn, R. L., Wolfe, D. M., Quinn, R. P., Snoek, J. D., and Rosenthal, R. A. (1964). Organizational Stress: Studies in Role Conflict and Ambiguity, Oxford: Wiley. Keil, M., Tan, B. C. Y., Wei, K., Saarinen, T., Tuunainen, V., and Wassenaar, A. (2000). A Cross-Cultural Study on Escalation of Commitment Behavior in Software Projects, MIS Quarterly, 24 (2), 299-324. Kim, D., and Benbasat, I. (2006). The effects of trust-assuring arguments on consumer trust in Internet stores: application of Toulmin's model of argumentation. Information Systems Research, 17(3), 286-300. Krasnova, H., Spiekermann, S., Koroleva, K., and Hildebrand, T. (2010). Online social networks: Why we disclose. Journal of Information Technology, 25(2), 109–125. Krasnova, H., Veltri, N. F., and Günther, O. (2012). Self-disclosure and Privacy Calculus on Social Networking Sites: The Role of Culture. Business & Information Systems Engineering, 4(3), 127-135. Lawler E. J., and Thye S. R. (1999). Bringing emotions into social exchange theory. Annual review of sociology, 25, 217–244. Li, Y., Kobsa, A., Knijnenburg, B. P., and Nguyen, M-H. C. (2017). Cross-Cultural Privacy Prediction. Proceedings on Privacy Enhancing Technologies, 3 (2), 1-19. Liu, Z., Min, Q., Zhai, Q., and Smyth, R. (2016). Self-disclosure in Chinese micro-blogging: A social exchange theory perspective. Information & Management, 53(1), 53-63. Liu, Z., Wang, X., and Liu, J. (2018a). How Digital Natives Make their Self-Disclosure Decisions: A Cross-Cultural Comparison. Information Technology & People. Liu, Z., Wang, X., Min, Q., and Li, W. (2018b). The Effect of Role Conflict on Self- disclosure in Social Network Sites: An Integrated Perspective of Boundary Regulation and Dual Process Model. Information Systems Journal. Lowry, P. B., Cao, J., and Everard, A. (2011). Privacy Concerns Versus Desire for Interpersonal Awareness in Driving the Use of Self-Disclosure Technologies: The Case of ACCEPTEDInstant Messaging in Two Cultures. Journal MANUSCRIPT of Management Information Systems, 27(4), 163–200. Malhotra, N. K., Kim, S. S., and Agarwal, J. (2004). Internet users' information privacy concerns (IUIPC): The construct, the scale, and a causal model. Information Systems Research, 15(4), 336-355. Markus, H., and Kitayama, S. (1991). Culture and the Self: Implications for Cognition, Emotion, and Motivation, Psychological Review, 98 (2), 224–253. 43

Mazin, S. (2014). Do You Have Too Many Facebook Friends? http://bigthink.com/praxis/do- you-have-too-many-facebook-friends. Accessed on 15-Dec-2016. Milberg, S. J., Burke, S. J., Smith, H. J., and Kallman, E. A. (1995). Values, personal information privacy, and regulatory approaches. Communications of the ACM, 38(12), 65- 74. Milberg, S. J., Smith, H. J., and Burke, S. J. (2000). Information Privacy: Corporate Management and National Regulation. Organization Science, 11(1), 35-57. Miltgen, C. L., and Peyrat-Guillard, D. (2014). Cultural and generational influences on privacy concerns: a qualitative study in seven European countries. European Journal of Information Systems, 23(2), 103-125. Nippert-Eng, C. (1996). Calendars and Keys: The Classification of “Home” and “Work”, Sociological Forum, 11(:3), 563-582. Park, C., Jun, J., and Lee T. (2013). Determinants of Use Intensity in Social Networking Sites: A Cross-cultural Study of Korea and USA. Proceedings of International Conference on e-Business, Reykjavik, Iceland. Park, C., Jun, J., Lee, T. (2015). Consumer characteristics and the use of social networking sites: A comparison between Korea and the US. International Marketing Review, 32(3/4), 414-437. Pentina, I., Zhang, L., Bata, H., and Chen, Y. (2016). Exploring privacy paradox in information-sensitive mobile app adoption: A cross-cultural comparison. Computers in Human Behavior, 65, 409-419. Peppas, S. C. (2001). Subcultural similarities and differences: An examination of US core values. Cross Cultural Management–An International Journal, 8(1), 59-70. Petronio, S.S. (2002). Boundaries of privacy: Dialectics of disclosure. Albany, NY: State University of New York Press. Podsakoff, P. M., MacKenzie, S. B., Lee, J.-Y., and Podsakoff, N. P. (2003). “Common Method Biases in Behavioral Research: A Critical Review of the Literature and Recommended Remedies.” Journal of applied psychology, 88 (5), 879-903. Posey, C., Lowry, P. B., Roberts, T. L., and Ellis, T. S. (2010). Proposing the online community self-disclosure model: the case of working professionals in France and the U.K. who use online communities. European Journal of Information Systems, 19(2), 181– 195. Postmes, T., Spears, R., and Lea, M. (1998). Breaching or building social boundaries? SIDE- effects of computer-mediated communication. Communication Research, 25(6), 689–715. Redpath, L., and Nielsen, M.O. (1997). A Comparison of Native Culture, Non-Native Culture and New Management Ideology. Canadian Journal of Administrative Sciences, 14(3), 327-339. Rizzo, J. R., House, R. J., and Lirtzman, S. I. (1970). Role Conflict and Ambiguity in Complex Organizations. Administrative Science Quarterly, 15 (2), 150-163. Rutner, P. S., Hardgrave, B. C., and McKnight, D. H. (2008). Emotional Dissonance and the Information Technology Professional, MIS Quarterly, 32 (3), 635-652. Schwartz, S. H. (1994). Beyond Individualism / Collectivism: New Cultural Dimensions of Values. In U. Kim, H. C. Triandis, Ç. Kâğitçibaşi, S. Choi, G. Yoon (Eds), Individualism ACCEPTEDand collectivism: Theory, method, and applications MANUSCRIPT (pp. 85-119). Thousand Oaks, CA: Sage. Sharma, S., and Crossler, R. E. (2014). Disclosing too much? Situational factors affecting information disclosure in social commerce environment. Electronic Commerce Research and Applications, 13(5), 305–319. Sheehan, K. B. (1999). An investigation of gender differences in on-line privacy concerns and resultant behaviors. Journal of Interactive Marketing, 13(4), 24-38. 44

Shih, H., Lai, K., and Cheng, T. C. E. (In Press). Constraint-based and dedication-based mechanisms for encouraging online self-disclosure: Is personalization the only thing that matters? European Journal of Information Systems. Smith, N. (2013). How much is a Facebook friend worth? $174.17. http://www.businessnewsdaily.com/4402-value-facebook-friend-marketing.html. Accessed on 15-Dec-2016. Smith, H. J., Dinev, T., and Xu, H. (2011) Information privacy research: An interdisciplinary review. MIS Quarterly, 35(4), 989–1015. Smith, H. J., Milberg, J. S., and Burke, J. S. (1996). Information privacy: Measuring individuals' concerns about organizational practices. MIS Quarterly, 20(2), 167-196. Srite, M., and Karahanna, E. (2006). The Role of Espoused National Cultural Values in Technology Acceptance. MIS Quarterly, 30(3), 679-704. Statista.com (2016). Daily time spent on social networking by internet users worldwide from 2012 to 2016 (in minutes). https://www.statista.com/statistics/433871/daily-social-media- usage-worldwide/. Accessed on 12-Dec-2016. Steenkamp, J. E. M., and Geyskens, I. (2006). How Country Characteristics Affect the Perceived Value of Web Sites. Journal of Marketing, 70(3), 136-150. Taylor, P. and Keeter, S. (Eds.) (2010). Millennials. Confident. Connected. Open to Change. The Pew Center, Washington D.C. Trepte, S., Reinecke, L., Ellison, N. B., Quiring, O., Yao, M. Z., and Ziegele, M. (2017). A Cross-Cultural Perspective on the Privacy Calculus. Social Media + Society, 3(1), 1–13. Trompenaars, F. (1993). Riding the Waves of Culture: Understanding Cultural Diversity in Business. London, UK: The Economist Books Turner, J. C. (1982). Towards a cognitive redefinition of the group. In H. Tajfel (Ed.), Social identity and intergroup relations (pp. 15-40). Cambridge, UK: Cambridge University Press. Utz, S., and Krämer, N. (2009). The privacy paradox on social network sites revisited: The role of individual characteristics and group norms. Cyberpsychology: Journal of Psychosocial Research on Cyberspace, 3(2), article 2. Wheeless, L.R. and Grotz, J. (1976). Conceptualization and measurement of reported self- disclosure, Human Communication Research, 2 (4), 338–346. Wang, X., and Li, Y. (2015). Users’ Satisfaction with Social Network Sites: A Self- Determination Perspective. Journal of Computer Information Systems, 56(1), 48-54. Wu, K., Huang, S. Y., Yen, D. C., and Popova, I. (2012). The effect of online privacy policy on consumer privacy concern and trust. Computers in Human Behavior, 28(3), 889-897. Xu, Heng. (2012). Reframing privacy 2.0 in online social networks. Journal of Constitutional Law, 14(4), 1077-1102. Xu, H., Dinev, T., Smith, J., and Hart, P. (2011). Information Privacy Concerns: Linking Individual Perceptions with Institutional Privacy Assurance. Journal of the Association for Information Systems, 12 (12), 798-824. Yang, K. C. C., and Kang, Y. (2015). Exploring Big Data and Privacy in Strategic Communication Campaigns: A Cross-Cultural Study of Mobile Social Media Users’ Daily Experiences. International Journal of Strategic Communication, 9(2), 87-101. Zhang,ACCEPTED N., Wang, C., and Xu, Y. (2011). Privacy MANUSCRIPT in Online Social Networks. Proceedings of International Conference on Information Systems. Zlatolas, L. N., Welzer, T., Heričko, M., and Hölbl, M. (2015). Privacy antecedents for SNS self-disclosure: The case of Facebook. Computers in Human Behavior, 45, 158–167.

Appendix A: A Brief Review of Privacy Literature Table A.1 Factors influencing self-disclosure on SNSs Attributes Definition/Description Sources Individual attributes Extroversion A positive emotion and tendency to seek Chen (2013) stimulation and others’ company Disclosure self-efficacy Perceived competency to perform information James et al. (In Press) disclosure Narcissism The feeling of being special and deserving much Utz and Krämer (2009) attention Privacy apathy A state of indifference toward privacy Sharma and Crossler (2014) Social interaction attributes Dependency The difficulty in seeking better relationships Shih et al. (In Press) Group privacy Unified privacy boundaries coordination De Wolf et al. (2014) management Network commonality The degree of network overlap Choi et al. (2015) Perceived enjoyment Perceptions of self-disclosure to be pleasant and Krasnova et al. (2012) entertaining Relationship Build up new connections and/or stay in touch with Chen and Sharma (2015); building/maintenance connections on SNSs Choi et al. (2015); James et al. (2015); Krasnova et al. (2010) Self-expression The willingness to have an online presence through James et al. (2015) disclosed information Severity of exposing The magnitude of exposing others through James et al. (In Press) others individuals’ activities on SNSs Social identity Personal similarity, social bonding, and value Shih et al. (In Press) connotation Social norms Norms regarding how personal profile should be Utz and Krämer (2009) setup Susceptibility of others to Perceived susceptibility of others to information James et al. (In Press) exposure exposure as a result of individuals’ activities on SNSs Switching cost The loss of loyalty benefits resulting from ending Shih et al. (In Press) current relationships Trust in others Beliefs about the trustworthiness of others with Krasnova et al. (2012); regard to disclosed information Shih et al. (In Press) Service attributes Critical mass Whether SNS has reached its critical mass to Chen (2013) achieve sustainable growth Cyber risk The likelihood of cyber-attacks because of using Chen and Sharma (2015) ACCEPTEDSNSs MANUSCRIPT Privacy policy A statement informing how personal information is Gerlach et al. (2015); handled Zlatolas et al. (2015) Trust in SNSs Beliefs about the trustworthiness of SNS providers Krasnova et al. (2012) External environment attributes Internet risk Individuals’ anxiety about using the Internet Chen (2013)

Table A. 2 Cross-cultural privacy studies Contex Country/Re Culture Analysis Study Main Findings t gion/City Dimension Approach Power Individualism index is distance; Culture as negatively related to privacy individuali Cecere given concerns; the index of 26 countries sm; et al. SNS (Index uncertainty avoidance, from Europe masculinity (2015) value is masculinity, and power ; used) distance is positively related uncertainty to privacy concerns. avoidance Collectivistic cultural leanings are negatively James et Culture as The U.S. and Individuali associated with perceived al. (In SNS independen South Korea sm severity of Facebook activity Press) t variables leading to the exposure of others' personal information. Individuali The negative effect of Krasnov The U.S. and sm; Culture as privacy concerns on self- a et al. SNS Germany uncertainty given disclosure is stronger in the (2012) avoidance German sample. The effect of privacy Park et concern on intensity of SNS The U.S. and Individuali Culture as al. SNS use is greater in the U.S. South Korea sm given (2013)ACCEPTED MANUSCRIPTsample than in the Korean sample. The effect of privacy Park et The U.S. and Individuali Culture as concern on SNS use is al. SNS South Korea sm given greater in the US sample (2015) than in the Korean sample.

Individualism and uncertainty avoidance have Germany, the negative effect on the Netherlands, Individuali Trepte et Culture as importance of avoiding the United sm; al. SNS independen privacy risks and social Kingdom, the uncertainty (2017) t variables gratification. Individualism U.S., and avoidance also has a direct effect on China willingness for uploading pictures. Culture as Yang Individualism index is given and The U.S. and Individuali positively correlated with SNS (Index Kang Taiwan sm privacy concerns and value is (2015) perceptions. used) Power distance; Uncertainty avoidance and individuali collectivism are positively Lowry et Instant Culture as The U.S. and sm; related to privacy concerns, al. Messag independen China masculinity whereas power distance is (2011) ing t variables ; negatively related to privacy uncertainty concerns. avoidance Culture as Collectivism is positively Posey et Online France and independen Individuali related to self-disclosure; no al. commu the United t variables sm statistical differences across (2010) nity Kingdom and as ACCEPTED MANUSCRIPTcountries. given Privacy concern does not Pentina Self- Mobile The U.S. and Culture as have significant effects on et al. direction; apps China given future use intentions of (2016) security private-information sensitive

apps in either the US or China.

Participants from the U.S. Chen et Mobile The U.S. and Individuali Culture as sample have higher levels of al. commer South Korea sm given privacy concerns than those (2013) ce from South Korean sample. The negative effect of Dinev et E- The U.S. and Individuali Culture as privacy concerns on e- al. commer Italy sm given commerce use is stronger for (2005) ce the American sample. People in more Steenka Privacy Culture as individualistic countries mp and in Individuali moderator 23 countries emphasize more on pleasure Geysken Website sm (in to privacy/security s (2006) s regression) protection. Power distance; Cultures have significant Culture as individuali effects on two dimensions of Bellman Informa moderator sm; information privacy (errors et al. tion 38 countries (to divide masculinity in databases and (2004) privacy pooled ; unauthorized secondary sample) uncertainty use). avoidance Internet users from cultures ACCEPTEDBangalore, MANUSCRIPT Individuali with high Cho et Informa Seoul, Culture as sm; individualism/uncertainty al. tion Singapore, independen uncertainty avoidance feel more (2009) privacy Sydney, and t variables avoidance concerned about online New York privacy.

Canada, China, Germany, the Culture as Informa U. S., the Individualism can better Li et al. Individuali moderator tion United predict privacy decisions (2017) sm (in privacy Kingdom, than country and language. regression) Sweden, Australia, and India Power Culture as distance; Milberg Informa given Cultures have no significant individuali et al. tion 30 countries (Index effects on sm; (1995) privacy value is information privacy concern. uncertainty used) avoidance Power distance; individuali Cultural values are Milberg Informa Culture as sm; positively associated with et al. tion 19 counties independen masculinity consumer information (2000) privacy t variables ; privacy concerns. uncertainty avoidance ACCEPTED MANUSCRIPT

Appendix B: Measurement Role Conflict (Rizzo et al., 1970; Rutner et al., 2008; Zhang et al., 2011) RC1 I post information that is apt to be accepted by some people but not accepted by others on this site. RC2 I often have to try to balance two or more conflicting activities on this site. RC3 I sometimes have to break a rule or norm to complete the thing I would like to do on this site. RC4 I frequently receive incompatible expectations from two or more parties on this site.

Role Overload (Zhang et al., 2011) RO1 I feel overburdened by different roles I have taken on this site. RO2 I have been given too much role expectation from other peers on this site. RO3 I have too many roles to play on this site. RO4 The amounts of roles I have to play interfere with the actual things I would like to do on this site. RO5 I feel I have too many roles to comfortably handle on this site. RO6 Different expectations from other peers make me too tired or irritable to participate in or enjoy the activities on this site.

Perceived Effectiveness of Privacy Settings (Xu et al., 2011) PE1 Social networking sites often provide certain settings to protect your privacy. PE2 I feel confident that these privacy settings reflect other peers’ commitments to protect my privacy. PE3 With these privacy settings, I believe that the content that I post will be kept private and confidential.

Privacy Control (Xu et al., 2011) PC1 I believe I can control the information posted on this site. PC2 I believe I have control over who can get access to my information posted on this site. PC3 I think I have control over what information is released on this site. PC4 I believe I have control over how my information is used by other peers on this site.

Privacy Risk (Xu et al., 2011) PR1 In general, it would be risky to post information on this site. PR2 There would be high potential for privacy loss associated with posting information on this site. PR3 My information would be inappropriately used by other peers. PR4 Posting information on this site would involve many unexpected problems.

Social Rewards (Jiang et al., 2013) SR1 I believed that posting information on this site would fulfill my social needs (for ACCEPTEDexample, companionship, approval, MANUSCRIPT acceptance, respect, status) in some way. SR2 I believed that posting information on this site would help me cultivate a good relationship with other peers. SR3 I believed that I could derive satisfaction from posting information on this site.

Intention to Self-disclose (Malhotra et al., 2004) 51

Specify the extent to which you would reveal (the information) through the Internet. SD1 (1) Unlikely/likely SD2 (2) Not probable/probable SD3 (3) Impossible/Possible SD4 (4) Unwilling/Willing

Disposition to Value Privacy (Xu et al., 2011) DTVP1 Compared to others, I am more sensitive about the way other peers use my information. DTVP2 To me, it is the most important thing to keep my information private. DTVP3 Compared to others, I tend to be more concerned about threats to my information privacy.

Previous Privacy Experiences (Xu et al., 2011) PEXP1 How often have you been a victim of what you felt was an improper invasion of privacy? PEXP2 How much have you heard or read during the past year about use and potential misuse of the information collected from the Internet? PEXP3 How often have you experienced incidents where your information on SNS was used by other peers without your authorization?

Trust toward SNSs (Krasnova et al., 2010) TSNS1 This site is open and receptive to the needs of its users. TSNS2 This site makes good-faith efforts to address user concerns. TSNS3 This site is also interested in the well-being of its users, not just its own. TSNS4 This site is honest in its dealing with users. TSNS5 This site keeps its commitments to its users. TSNS6 This site is trustworthy.

Trust toward other SNS users (Krasnova et al., 2010) TUSER1 Other peers on the site will do their best to help me. TUSER2 Other peers on the site do care about the well-being of others. TUSER3 Other peers on the site are open and receptive to the needs of each other. TUSER4 Other peers on the site are honest in dealing with each other. TUSER5 Other peers on the site keep their promises. TUSER6 Other peers on the site are trustworthy.

Power Distance (Chen and Zahedi, 2016) PD1 People in higher positions should not delegate important tasks to people in lower positions. PD2 People in higher positions should not ask the opinions of people in lower positions too frequently. PD3 People in higher positions should avoid social interaction with people in lower ACCEPTEDpositions. MANUSCRIPT

Uncertainty Avoidance (Chen and Zahedi, 2016) UA1 It is important to have instructions spelled out in detail so that I always know what I am expected to do. UA2 It is important to closely follow instructions and procedures. 52

UA3 Standardized work procedures are helpful. UA4 Compared to having general directions, having detailed instructions on how to do my job is important. UA5 Compared to an ambiguous environment that allows for personal innovation, having standardized job description is important.

Collectivism (Chen and Zahedi, 2016) COL1 Individuals should sacrifice self-interest for the group. COL2 Individual should stick with the group even through difficulties. COL3 Group welfare is more important than individual rewards. COL4 Group success is more important than individal success. COL5 Group loyalty should be encouraged even if individual goals suffer.

ACCEPTED MANUSCRIPT

Appendix C: Measurement Assessment

Table C.1. Descriptive statistics of items US China Construct Items Mean SD Loading Mean SD Loading DTVP1 4.46 2.57 .89 5.25 1.79 .78 Disposition to DTVP2 5.43 1.98 .75 5.92 1.55 .88 Value Privacy DTVP3 4.84 2.25 .91 5.52 2.09 .85 GNC 5.35 1.65 .93 5.00 1.30 .94 Group Norms GNS 5.51 1.81 .89 5.23 1.42 .91 PC1 4.91 2.74 .83 5.26 1.67 .71 PC2 4.79 2.69 .89 5.364 1.67 .78 Privacy Control PC3 4.73 2.49 .90 4.85 1.97 .83 PC4 4.49 2.86 .88 4.44 2.17 .72 PE1 4.87 2.10 .89 4.94 1.96 .78 Perceived PE2 4.67 2.46 .91 5.37 1.62 .90 Effectiveness PE3 4.91 2.10 .92 5.43 1.72 .87 PR1 4.17 2.88 .88 4.91 2.01 .86 PR2 4.27 2.64 .85 4.71 2.32 .89 Privacy Risk PR3 3.87 2.52 .88 5.08 2.00 .87 PR4 3.80 2.82 .89 4.90 1.92 .83 RC1 3.95 3.90 .76 5.08 1.72 .65 RC2 3.29 3.82 .87 4.29 2.34 .83 Role Conflict RC3 2.86 3.68 .88 3.96 2.63 .77 RC4 3.10 3.49 .87 4.18 2.32 .75 RO1 2.80 3.20 .90 3.81 2.42 .79 RO2 2.74 3.10 .92 3.91 2.08 .76 RO3 2.63 2.99 .93 3.90 2.21 .70 Role Overload RO4 2.82 3.56 .91 3.28 2.18 .84 RO5 2.78 3.45 .92 3.36 1.99 .82 RO6 2.83 3.38 .89 3.60 2.72 .78 SR1 4.30 2.74 .86 5.34 1.19 .85 Social Rewards SR2 4.65 1.98 .90 5.54 1.05 .85 SR3 4.68 2.22 .90 5.44 1.26 .84 SD1 4.39 2.70 .93 5.30 1.48 .77 Intention to SD2 4.34 2.63 .95 4.59 1.97 .91 ACCEPTEDSelf-disclose SD3 4.43 MANUSCRIPT2.79 .93 4.54 1.97 .92 SD4 4.35 2.75 .92 4.35 2.55 .88 PD1 3.79 3.06 .78 2.95 2.57 .57 Power Distance PD2 3.22 3.12 .90 2.673 2.15 .56 PD3 2.99 3.56 .86 2.96 2.20 .99

UA1 5.11 2.02 .78 5.29 1.86 .69 UA2 5.42 1.71 .81 5.67 1.43 .83 Uncertainty UA3 5.42 1.67 .80 5.75 1.36 .82 Avoidance UA4 5.26 1.68 .84 5.62 1.45 .83 UA5 4.92 1.69 .75 5.42 1.74 .80 COL1 3.98 2.71 .80 4.37 2.10 .79 COL2 4.63 1.99 .75 4.74 2.17 .82 Collectivism COL3 4.44 2.12 .81 4.90 2.32 .87 COL4 4.45 2.23 .85 4.90 2.35 .85 COL5 4.33 2.35 .84 4.93 2.00 .77 Previous PEXP1 2.58 2.49 .94 3.74 1.37 .79 Privacy PEXP2 4.00 2.51 .60 4.70 1.47 .73 Experiences PEXP3 2.32 2.67 .94 3.50 2.06 .82 TSNS1 5.14 1.27 .81 5.25 1.14 .80 TSNS2 5.08 1.30 .86 5.23 1.21 .86 Trust toward TSNS3 5.01 1.39 .91 5.31 1.24 .83 SNSs TSNS4 4.96 1.35 .92 5.30 1.19 .88 TSNS5 4.95 1.35 .89 5.41 1.17 .88 TSNS6 5.02 1.36 .87 5.56 1.13 .84 TUSER1 4.97 1.30 .85 5.36 1.12 .80 TUSER2 5.01 1.27 .82 4.65 1.25 .77 Trust toward TUSER3 4.98 1.24 .90 4.90 1.20 .86 Other Users TUSER4 4.87 1.31 .89 4.94 1.28 .90 TUSER5 4.78 1.32 .89 4.91 1.25 .88 TUSER6 4.88 1.30 .88 4.97 1.26 .85

ACCEPTED MANUSCRIPT

Appendix D: The Moderating Effect of Culture on Self-disclosure

In this appendix, we discuss and assess the moderating effect of culture on the

relationship between boundary rule formation and intention to self-disclose. We first

discuss how each relationship is moderated by different culture dimensions and then

conduct our analysis by comparing the structural model between the U.S. and China

samples as well as across subsamples divided by espoused cultural dimensions.

Boundary Rule Formation and Self-Disclosure

Individuals with high collectivism (i.e., Chinese) usually emphasize more social

relationships with others. Therefore, they are more likely to value the social rewards

derived from interacting with others on SNSs and to maintain their social

relationships by self-disclosing. Therefore, we propose that the effect of social

rewards on intention to self-disclose is stronger in cultures with high collectivism

(China > US).

Further, individuals with high uncertainty avoidance (i.e., Americans) are less

tolerant to risk. Therefore, they are more likely to perceive a more severe risk to their

privacy. As a consequence, they tend to avoid privacy risks by not disclosing their

information. Therefore, we propose that the effect of privacy risk on intention to self-

disclose is stronger in cultures with high uncertainty avoidance (US > China).

Finally, uncertainty avoidance may also influence the effect of disposition to value

ACCEPTEDprivacy. Individuals with high uncertainty MANUSCRIPTavoidance (i.e., Americans) are more

concerned with potential risks. Therefore, given the same level of disposition to value

privacy, such individuals may guard their privacy boundaries even more strictly. In 56

such a context, they feel even more concerned about their privacy and more negative

toward self-disclosure. Therefore, we propose that the effect of disposition to value

privacy on intention to self-disclose is stronger in cultures with high uncertainty

avoidance (US > China)

Data Analysis and Results

We first compare the structural model between the US and China sample (Table

D.1). Only the relationship disposition to value privacy → self-disclosure is consistent

with our prediction. Table D.1. The structural model between the US and China Hypothesis US China T-value Diff. Supported? Sig. H1: Social rewards → Self-

.47*** .33*** 8.44 >*** No disclosure H2: Privacy risk →Self- .05 -.25*** 7.01 sd No disclosure

Constructs H4a: Disposition to value Theoretical Theoretical -.17** -.10* -4.31 <*** Yes privacy →Self-disclosure Age →Self-disclosure -.07 -.09* Gender →Self-disclosure -.11*** .06 Number of contact types →

.03 .07 Self-disclosure Previous experiences → .11* .05 Self-disclosure

Covariates Trust toward SNSs →Self- .11 .13* disclosure Trust toward other users → .14* .05 Self-disclosure R2 .46 .29 * p < .05, ** p < .01, *** p < .001, sd = structurally different (one path is significant and the other is not), Diff. Sig. = Different Significantly, Supported? = Is the prediction supported?

ACCEPTEDWe then compare the structural model acrossMANUSCRIPT subsamples divided by espoused

cultural dimensions (Table D.2). Uncertainty avoidance moderates the relationship

privacy risk → self-disclosure, where the effect of privacy is stronger under low

uncertainty avoidance. Collectivism moderates the relationship social rewards → self-

disclosure and privacy risk → self-disclosure. Here the effect of social rewards is

stronger under low collectivism, whereas, the effect of privacy risk is stronger under

high collectivism. Table D.2. Post hoc analysis of the moderation of espoused cultural dimensions Hypothesis Uncertainty Avoidance Collectivism T- Dif T- Diff Low High Val f. Low High Valu . ue Sig. e Sig. H1: Social rewards → .45** .43* .52** .33* >** 1.20 ns 11.93 Self-disclosure * ** * ** * H2: Privacy risk →Self- - - -.03 sd -.06 -.11* 3.47 sd disclosure .17** 0.84 H4a: Disposition to value - -

Constructs

Theoretical Theoretical - → .16* 1.63 ns .17** -.14* -1.37 ns privacy Self-disclosure .13** * * Age →Self-disclosure -.06 -.05 -.04 -.07 Gender →Self-disclosure -.01 -.01 -.01 -.02 Number of contact types

.05 .03 .06 .04 →Self-disclosure Previous experiences → .11 .08 .06 .13 Self-disclosure

Covariates Trust toward SNSs → .12* .08 .10 .10 Self-disclosure Trust toward other users .12* .10 .13 .10 →Self-disclosure R2 .39 .33 .42 .29 * p < .05, ** p < .01, *** p < .001, ns = not significant, sd = structurally different (one path is significant and the other is not), Diff. Sig. = Different Significantly

Discussions

In our comparison, the effect of social rewards on self-disclosure is stronger for

Americans. Our post hoc analysis shows that the effect of social rewards is stronger

ACCEPTEDunder low collectivism culture, not consistent MANUSCRIPT with our prediction.

Privacy risk is only significantly related to intention to self-disclose for Chinese.

Although these results do not support our hypotheses, they are consistent with the 58

findings from Chen and Zahedi (2016). Specifically, Chen and Zahedi (2016) argue

that because the Chinese government presently cannot fully control the Internet to

preserve social harmony, Chinese people have more exposure to negative information

and are more sensitive to the potential risks.

The effect of disposition to value privacy on intention to self-disclose is stronger

for Americans than for Chinese. Post hoc analysis also shows that disposition to value

privacy has a stronger effect on intention to self-disclose under high uncertainty

avoidance, consistent with our prediction.

Overall, these results show that culture cannot fully explain the relationship

between boundary rule formation and intention to self-disclose, and there may be

other contextual factors jointly influencing individuals’ self-disclosure decisions.

Boundary Coordination Boundary Rule Formation

Group Norms H4a H5a U

Perceived Effectiveness H4b H6a U>C of Privacy Settings Cost-Benefit Assessment

H6b U>C Social Rewards

Boundary Turbulence H3a Information Disclosure H7 U>C Role Conflict Privacy Control Intention to self-disclose

H3b

Role overload Privacy Risk ACCEPTEDH8 U>C MANUSCRIPTH4c H2

US (U) vs. China (C)

Figure 1. Research Model 59

ACCEPTED MANUSCRIPT