Privacy in Mediated and Nonmediated Interpersonal Communication: How Subjective Concepts and Situational Perceptions Influence Behaviors

Privacy as a Theoretical Concept in Communication

Communication scholars most often refer to Westin’s (1967) and Altman’s (1975) theories as most important for laying the groundwork for the current understanding of privacy (for an overview, see Margulis, 2011). Westin (1967) defined privacy as “the claim of individuals, groups, or institutions to determine for themselves when, how, and to what extent information about themselves is communicated to others” (p. 5). People achieve this by voluntarily and temporarily withdrawing from social interactions. In modern societies, privacy is necessary for the individual to maintain personal autonomy, emotional release, and self-evaluation, as well as limited and protected communication. According to Westin, these functions can be performed in four states of privacy: solitude, intimacy, anonymity, and reserve. The individual desire for privacy competes with other social and psychological needs and is shaped by the social and physical environment. As it changes constantly, an individual may perceive the given level of privacy as too much or too little. Although Westin does not explicitly discuss privacy perceptions, his work nonetheless suggests that the subjective perception of a given level of privacy must be regarded as a requirement for the personal adjustment process aimed at achieving a sufficient level of privacy.

Whereas Westin regarded privacy as resulting from withdrawal, Altman (1975) defined privacy as “selective control of access to the self” (p. 18). He used privacy synonymously with interpersonal boundary control through which individuals or larger social units aim to achieve a temporarily desired level of interpersonal contact. He emphasized the dialectical nature of the process that involves the regulation of input and output. Thus, Altman’s concept of privacy is not only a process of withdrawing or a state of limited access. Privacy is rather an interplay of opposing forces, also including the desire to be accessible to others. One of Altman’s major claims was that individuals compare their desired with their achieved level of privacy. He also called the two levels the ideal and the outcome, respectively (Altman, 1975). The ideal level of privacy is defined by the desired level of openness or closeness in a situation. Similar to Westin, Altman suggested that a boundary regulation process takes place whenever the achieved level of privacy is perceived as failing to match the desired level of privacy. Although Altman likewise did not explicitly refer to privacy perceptions as a prerequisite for privacy regulation processes, his concept implicitly necessitates that individuals subjectively assess the given level of privacy in any situation to evaluate it against their desired level of privacy.

Burgoon (1982) built on Westin’s (1967) and Altman’s (1975) work to answer the question of whether there are “different kinds of dimensions of privacy that have implications for communication” (p. 206). She suggested four interrelated dimensions of privacy: the ability to control physical, interactional, psychological, and informational access to oneself or to one’s group. On the physical dimension, individuals regulate the degree of surveillance as well as physical access to their personal space. The interactional dimension of privacy encompasses autonomous engagement in or withdrawal from social encounters. The psychological dimension involves individuals’ ability to prevent intrusions upon their cognitions and feelings. Finally, informational privacy is the ability to control the gathering and disseminating of information about the self (Burgoon et al., 1989). Burgoon (1982) emphasized the subjective nature of privacy and the role of an individual privacy perception as she stated that the “degree of privacy in any given situation is therefore dependent on each individual’s interpretation of the situation” (p. 211).

With her communication privacy management (CPM) theory, Petronio (2002) further refined Altman’s (1975) privacy concept of selective control. CPM offers a rule-based conceptualization of the mental calculus underlying the decision to hide or release private information. Central to the CPM is the ownership principle, which postulates that individuals see themselves as the owners of their private information from which they derive a right to control the flow of it. The sharing of private information is based on privacy rules that are driven by expected benefits and costs. Once an individual grants others access to private information, they become co-owners. Collective privacy rules, which are mutually agreed upon by the original owner and the co-owners, regulate how they further share or keep the disclosed information. According to Petronio, co-owning individuals manage privacy boundaries by applying permeability rules, which determine how much is shared; linkage rules, which determine who gets to know the information; and ownership rules, which determine how much control co-owners have over the private information. If co-owners fail to coordinate the privacy rules, CPM theory predicts boundary turbulence (i.e., privacy violations), which requires a recalibration of the established rules. Privacy perceptions, although not explicitly referred to as such, are relevant for privacy rule development. The perception of the physical setting and the social environment as safe and trustworthy are relevant criteria for rules of boundary regulation (Petronio, 2002).

Inherent to all the theoretical privacy concepts discussed here is the idea that individuals or larger social units achieve privacy by controlling access to information. Depending on the perceived level of privacy, individuals apply rules or choose behaviors in order to achieve and maintain their desired level of privacy. Although only Burgoon (1982) explicitly included the perception of the level of privacy in her concept, subjectively perceiving privacy and reacting accordingly must be regarded as the fundamental mechanisms of all privacy concepts.

Social media challenge this privacy mechanism in two ways. On one hand, unknown and invisible privacy risks impede a valid perception of the actual (or achieved) privacy level in communication settings. On the other hand, the infrastructure and properties of digital media limit individuals’ capabilities to control the flow of personal information efficiently. In social media, the behaviors used to implement a desired level of privacy are somewhat limited. The changes that new communication media have brought for privacy management in everyday life demand a theoretical examination of privacy that takes these altered circumstances into consideration (Papathanassopoulos, 2015).

Theoretical Approaches to Privacy in a Social Media Era

Since social media’s popularity has boomed, the means for interpersonal communication and privacy and data protection have re-emerged as topics of interest. Various applications collect, store, and process a variety of data about users, often without their awareness or the opportunity to opt out. Consequently, communication scholars have investigated how individuals manage their privacy in social media (boyd, 2008a; boyd & Hargittai, 2010; Gross & Acquisti, 2005; Joinson & Paine, 2007; Masur, 2018; Papacharissi, 2010; Trepte, Dienlin, & Reinecke, 2014; Trepte & Reinecke, 2011; Tufekci, 2008). Such research has given first insights into how the digital environment, social constellations, and individual needs challenge privacy perceptions and privacy regulation in social media.

On the basis of these previous studies, we found three aspects that most crucially influence and complement privacy theories. First, in social media, the concept of audience is different than in face-to-face communication. Social media audiences are potentially large and unknown and cannot be physically perceived (Litt, 2012). Beyond other users, the audiences include service providers, surveillance agencies, and other third parties that typically do not exist in nonmediated communication settings. In most privacy theories, communication partners are usually referred to as others. Altman (1975) summarized that privacy involves “individuals, families, mixed and homogeneous sex groups, and so on” (p. 11). Similarly, in Westin’s (1967) and Petronio’s (2002) privacy theories, it is assumed that individuals have a general idea of who the others are and that one can communicate with them. While communicating in social media, potentially unknown audiences may impede an accurate evaluation of the given level of privacy.

Second, digital data are persistent, replicable, searchable, and scalable, and information addressing a defined circle of recipients might spread further than expected or intended (boyd, 2011; Palen & Dourish, 2003). Most of the privacy theories consider the transfer of information: Petronio (2002), for example, argued that information transfer is negotiated between co-owners. The properties of social media undermine the rules established in negotiations because users sometimes unintentionally spread co-owned information beyond the agreed-upon boundaries. Social media are designed to potentiate information sharing. Thus, a heightened scalability has to be taken into account. Yet, not every user is aware of these characteristics or the fact that he or she might come along with privacy risks. Thus, users may have an inaccurate perception of the actual level of privacy in social media.

Third, multiple physical contexts with distinct behavioral norms that used to be separated by temporal, spatial, and social boundaries converge in social media. This collapsing of contexts causes privacy tensions, especially in one-to-many communication (Binder, Howes, & Sutcliffe, 2009; boyd, 2008b, 2011; Palen & Dourish, 2003). This is a fundamental challenge for privacy management in social media for which the traditional theories of privacy provide an adequate theoretical framework only in part.

In sum, the properties of social media require privacy regulations that are vastly different from offline settings in terms of audiences, access, and boundary control. Marwick and boyd (2014) found that social media users need to establish collective practices in order to safeguard their privacy. On the basis of these observations, they proposed “a model of privacy that is networked” (p. 12). This model requires an understanding of the technological and social peculiarities of social media contexts as well as shared social norms over information sharing. Although we agree that a contemporary understanding has to regard privacy as a collective endeavor, we suggest that a better understanding of how it is established in concrete situations is needed. As supposed by Altman (1974), Westin (1967), Burgoon (1982), and Petronio (2002), the properties of digital communication and online audiences challenge users’ ability to properly perceive privacy levels and effectively control the flow of private information. Thus, there is a need to adapt the conceptualization so that privacy perceptions and behavior can be investigated adequately in a digitalized world.

Gaining a refined understanding of laypeople’s privacy concepts is a prerequisite for investigating their subjective perception of privacy levels in different communication settings. Knowing what privacy means to them allows us to develop a refined understanding of the cues and conditions that determine the perception of a given level of privacy. Hence, we explored individual privacy concepts in qualitative interviews and asked the following research question:

How Individuals Perceive Privacy in Mediated and Nonmediated Settings

So far, scholars have investigated people’s privacy concerns, attitudes, and behaviors (e.g., Barnes, 2006; Taddicken, 2014), but there is hardly any research on how people perceive different levels of privacy in mediated and nonmediated communication settings. To distinguish privacy perceptions from related concepts such as privacy concerns, we define privacy perceptions as individuals’ situational experiences with and assessments of their given privacy level. This implies that privacy perceptions can differ greatly among individuals (Dienlin, 2014). The perception of the given level of privacy is necessary in order to decide whether it needs to be adjusted or whether it is already optimal (Altman, 1975; Burgoon, 1982).

In order to determine how people experience and assess privacy perceptions, we are interested in the situational cues and conditions that determine privacy perceptions in various mediated and nonmediated communication settings. Burgoon (1982) suggested several factors for distinguishing privacy perceptions. According to her, the perception of physical privacy depends on the extent to which an individual can control access to the surrounding space and on the freedom of surveillance and the number of sensory channels through which access to the self is possible. A space is perceived as more private when it allows for seclusion and when the probability of intrusion is low. Factors determining perceptions of interactional privacy include the degree of control over communication partners and the frequency, length, and content of interactions. The perception of psychological privacy depends on the perceived freedom from the influence of others on cognitions and emotions and the ability to conceal thoughts and feelings. When both are low, one does not perceive a high level of privacy. Finally, the perception of informational privacy depends on the degree of control over the initial release of information and its subsequent distribution and use. Additional factors refer to the amount of information known by others and the number of people who have access to it. As Burgoon had defined privacy perception two decades before social media emerged, her theory does not yet include factors that might explain differences between the perception of privacy in mediated and nonmediated communication settings. Nonetheless, her abstract systematization of situational cues that might lead to a higher or lower perceived level of privacy may still be useful in understanding the potential of social media properties for shaping users’ privacy perceptions.

Although the term privacy perception has sometimes been used interchangeably with privacy concerns or privacy attitudes in the literature, we want to emphasize that we consider them to be different concepts. The focus of this work is on privacy perceptions in terms of a situational experience and the assessment of privacy in mediated and nonmediated communication settings. Thus, we do not focus on overall attitudes or concerns. Moreover, no systematic comparison of privacy perceptions in mediated and nonmediated communication exists so far. We therefore asked the following research question:

How Individual Privacy Perceptions Influence Self-Disclosure

Privacy perceptions as defined in the preceding section are important for assessing whether the given external conditions allow for the disclosure of private information (Masur, 2018). Self-disclosure has generally been defined as sharing information about the self with other humans (Cozby, 1973; Wheeless & Grotz, 1976). An individual’s decision to disclose depends on the perception that one is in control of the kind of information that is shared and with whom. If an individual perceives that a communication setting is not private enough for sharing personal information, people will regulate their privacy by not divulging personal details (Derlega, Metts, Petronio, & Margulis, 1993). Early evidence was provided by Fidler and Kleinknecht (1977), who interviewed female college students on stigmatizing information. They found that they received more responses to the most sensitive questions when the interview technique guaranteed a high level of privacy.

The emergence of computer-mediated communication in chatrooms and online discussion forums has stimulated research on online self-disclosure patterns. A major finding was that study participants who perceived privacy in an anonymous setting disclosed significantly more sensitive information than they did in similar offline interactions (Joinson, 2001; Tidwell & Walther, 2002).

When social media, especially SNSs, replaced anonymous means of online communication, scholars focused on the influence of privacy concerns—a concept related to privacy perceptions—on self-disclosure (Dienlin & Trepte, 2015; Krasnova, Spiekermann, Koroleva, & Hildebrand, 2010; Taddei & Contena, 2013; Taddicken, 2014; Tufekci, 2008). Masur and Scharkow (2016) found that users tended to refrain from sharing information they subjectively perceived as private. Accordingly, posting particularly intimate information publicly was evaluated as inappropriate, a finding that was confirmed by a content analysis of status updates on SNSs (Bazarova, 2012). Studies have found that sensitive information tends to be shared in communication settings that are perceived as allowing a higher level of privacy, for example, email or instant messaging (Bazarova & Choi, 2014; Burkell, Fortier, Wong, & Simpson, 2014; McLaughlin & Vitak, 2012; Utz, 2015). Based on an experience sampling study with N = 164 smartphone users who engaged in N = 1,104 disclosure events, Masur (2018) showed that the environmental circumstances within a communication application allow for the prediction of the depth of self-disclosure. More specifically, the findings revealed that the depth of self-disclosure increased in situations in which the audience size was small (e.g., IM conversations) and potential respondents were considered as trustworthy and psychologically close. Similarly, Frye and Dornisch (2010) measured perceived privacy in terms of the perception of not being overheard or read by unintended audiences and the willingness to self-disclose on 32 topics across 10 communication settings. The results again showed that participants were more willing to self-disclose information when using a communication tool that was perceived to provide a high level of privacy.

So far, we have gained insight into how people adjust communication content with respect to public and private means of communicating on SNSs, and there is convincing evidence that privacy perceptions and self-disclosure are positively correlated for various communication settings. For our study, we followed up on Frye and Dornisch’s (2010) systematic approach and explored how privacy perceptions in different mediated and nonmediated communication settings influence the disclosure of private information. We specifically focused on disclosures that the individual refers to as private. As privacy attributions for topics or types of information vary considerably between individuals (Masur & Scharkow, 2016), we did not classify information as private a priori but left it open to our participants to come up with an experience involving private disclosures. Thus, we sought to understand how privacy perceptions impact the kind of private information users disclose and how they communicate about it.

Method

Results

Discussion

With the work presented in this article, we pursued the aim of increasing the understanding of privacy perceptions, a concept that is implicitly part of all academic privacy theories but has not received much attention in research on online privacy. A central finding of our study is that people’s privacy perceptions in social media are still determined by the same types of cues that Burgoon (1982), for example, had already systematized much earlier but that the accessibility of cues has changed tremendously. For example, whereas access to private information is protected by walls and doors in face-to-face situations, it is now protected by privacy settings or even the infrastructure of the respective online environment. Physical barriers are easy to perceive and can thus be evaluated with regard to their potential to provide privacy. By contrast, digital barriers cannot be perceived in the same way. Individuals have to possess a certain amount of knowledge and access this knowledge in relevant communication situations in order to adequately evaluate the given level of privacy. Recent work on the role of online privacy literacy provides further evidence for such a claim (Masur, Teutsch, & Trepte, 2017; Park, 2013). It is hence not surprising that our study revealed that trust is a central factor for both, the individual perception of privacy levels in different mediated and nonmediated communication settings and for laypeople’s subjective privacy concepts. Trust becomes a “risk mitigator” in environments in which the actual evaluation of its specific properties becomes challenging. Hence, we suggest that trust should be at the heart of a renewed privacy concept that accounts for the achievement of privacy in social media communication.

On the basis of the interviews, we further identified five communication settings in which private information is generally shared but that differs greatly regarding the perceived level of privacy: (a) face-to-face communication in dyads or small groups, (b) phone calls or video chats in dyads, (c) face-to-face communication in large or heterogeneous groups, (d) online messages such as email or IMs, and (e) semipublic or public online interactions on SNSs or online discussion boards. We found that the perceived level of privacy in these communication settings depended on a number of criteria that refer to social boundaries (audience size, interpersonal trust), physical boundaries (the existence of environmental artifacts for the protection of privacy), and the nature of communication (nonverbal, paraverbal, immediacy). Although subjective privacy perceptions varied from person to person, it is remarkable that participants nonetheless expressed consistent privacy perceptions across four of the communication settings. We found disagreement only with regard to the perception of the privacy level in IM communication.

As a main finding of our study, we hence suggest that privacy perceptions in small-group face-to-face encounters, phone calls and video chats, as well as encounters in large heterogeneous groups and public communication in social media (i.e., the four settings in which we found consistency) have undergone a process of norming. We believe people have developed and now share common norms that guide and regulate an appropriate flow of information in these settings. The first two communication settings are commonly perceived as private, resulting in the expectation that private matters can be openly discussed. The latter two communication settings are not perceived as private, and private disclosures are consequently avoided. The norming of privacy perceptions in these settings may have evolved from typical experiences, from similar socialization processes, and, in the case of public communication on SNSs, from mass media’s consistent framing of SNSs as a risk for security and informational self-determination (Teutsch & Niemann, 2016).

However, with respect to messenger communication, privacy perceptions were divergent: Some of our interviewees perceived this communication setting as private, some did not see it as suitable for private communication, and others reported ambivalence. This divergence also held true for the behavior shown in these settings: Not all people who perceived messenger communication as ambiguous or as not private avoided sharing private matters. Hence, privacy perceptions in messenger communication do not yet conform to commonly shared norms. We argue that they are still in a phase of storming. Communicating via IMs is still relatively new, yet for many people, it offers a convenient and exciting way to interact with others. Users are exploring this new realm of online communication and have not yet formed a set of privacy rules. This may be due to the nature of messenger communication. On one hand, most people perceive the usually dyadic or small-group conversations as private and protected. The audience size seems manageable, and social boundaries of privacy seem to be under control. On the other hand, users are aware of the risks evoked by the persistence and replicability of digital communication and the surveillance practices of service providers. In addition, privacy and data protection practices of the most popular messenger app WhatsApp change frequently. For example, the originally independent company WhatsApp, Inc., was acquired by Facebook in 2014. Although all WhatsApp conversations are end-to-end encrypted, in August 2016, WhatsApp announced that it would start to connect the service with Facebook and share account information. Consequently, users are uncertain about whether their communication will be protected and secure. As the massive use of messenger apps is a relatively new communication habit, we assume that over time, users will have many experiences and negotiations. Hence, privacy perceptions will also undergo a process of norming.

The challenge and uncertainty of forming a reliable perception of the level of privacy in mediated communication were also reflected in our participants’ subjective privacy concepts. On one hand, they referred to aspects of privacy that have been described in academic privacy theories—namely, self-determination, access regulation, and the management of social and physical boundaries—but on the other hand, they added new aspects to these privacy concepts that are crucially influenced by social media use. With regard to subjective concepts consistent with academic approaches, participants often claimed that they wanted to determine for themselves how other people see them or think about them—resembling Westin’s (1967) concept of privacy. They further expressed that privacy is about managing access to the self within certain social and physical boundaries. This idea of privacy as access control was clearly reflected in Altman’s (1975) theory of privacy and in Petronio’s (2002) CPM. Physical and social boundaries were outlined by Burgoon (1982) and later Petronio (2002) with an emphasis on how these boundaries are negotiated and managed. A crucial enhancement of our academic understanding of privacy in the digital age is interpersonal trust, which our interviewees repeatedly referred to as an important requirement for perceived privacy. In the age of social media, when controlling the flow of information is not feasible, trust seems to have gained enormous importance. The seminal theories of privacy have not yet explicitly incorporated the concept of trust. However, more current conceptualizations, and especially empirical studies on online privacy, refer to trust as an important variable (Krasnova et al., 2010; Masur, 2018; Miltgen & Smith, 2015; Taddei & Contena, 2013). In this sense, interpersonal perceptions of trustworthiness determine the level of privacy that is experienced. In support of this, Masur’s (2018) analysis of varying disclosure situations showed that depth of self-disclosure was indeed positively correlated to interpersonal assessments including interpersonal trust and psychological closeness. Despite such interpersonal evaluations, however, the risk that even trusted people might divulge what they should have kept to themselves remains. This risk seems to be particularly problematic if private communication is mediated. By sharing private information via messenger apps, the senders can never be entirely sure that the addressees are alone or that they are hiding their messages from unintended readers.

In sum, trust—a concept that was widely overlooked in early privacy research but is being considered more often in current studies—becomes more and more important for the definition of privacy. It is best understood as a boundary condition of privacy: Only if I trust the recipients of my disclosures will I experience privacy. For a novel privacy concept, not only is the interpersonal trust that was emphasized by our participants crucial but so is trust in online service providers and in the security of digital communication applications. As individual access regulation and control over personal data are hindered in social media, opting for trustworthy communication channels is a meaningful way to achieve and maintain privacy. Analogous to the above framed boundary condition, people experience privacy only if they consider a communication setting to be trustworthy, which involves trust in technology and service providers in social media communication. However, trust is not always well informed but may rely on heuristic information processing (Joeckel, Dogruel, & Bowman, 2016). It is thus inextricably linked with privacy perceptions and privacy behavior, and should always be considered a theoretical component of privacy.

Supplemental Material