Dimensionalizing privacy to advance the study of digital disempowerment

Abstract

In this essay, we call attention to privacy as the foundational construct that underpins digital disempowerment. We argue that to better understand the processes of disempowerment, scholars must critically engage with the dimensionality of privacy conceptualizations and privacy-dependent constructs such as privacy concerns and privacy-protecting behavior, and the way in which these are measured. We focus on privacy's horizontal and vertical dimensions as a way to offer a more nuanced understanding of power in computationally mediated environments and potentially enable a more refined and meaningful understanding of privacy resignation and disengagement.

This article is a part of special theme on Digital Resignation and Privacy Cynicism. To see a full list of all articles in this special theme, please click here: https://journals.sagepub.com/page/bds/digitalresignationandprivacycynicism

For nearly two decades, researchers have sought to explain the so-called privacy paradox, or the apparent disconnect between an individual's privacy concerns and their privacy-protecting behaviors (Barnes, 2006). Recent work highlights that a sense of disempowerment may be another explanation for this phenomenon. Labeled as privacy fatigue (Choi et al., 2018), privacy cynicism (Hargittai and Marwick, 2016; Hoffmann et al., 2016), or digital resignation (Bagger, 2021; Draper and Turow, 2019), this body of work foregrounds the perceived lack of agency individuals feel in their aspiration for greater digital privacy. Further, there is a growing recognition that such sense of disempowerment takes on different shapes and magnitudes as a function of context (Nissenbaum, 2010) and one's position within the power structure (Wu et al., 2019). In this essay, we call attention to how research conceptualizes and measures privacy as the foundational construct that underpins observations about digital resignation, its contexts, predictors, and its social and political importance. Following the lead of Raynes-Goldie (2010) and others (e.g. Bazarova and Masur, 2020; Lutz and Strathoff, 2013; Quinn et al., 2019), we argue for the deconstruction of privacy's conceptualizations to further distinguish its social and, especially, its institutional dimensions. Horizontally oriented privacy refers to social privacy relationships among individuals or groups (Raynes-Goldie, 2010), while vertically oriented privacy refers to the privacy relationships between an individual and institutions such as schools, government, or corporate platform sponsors (Bazarova and Masur, 2020). Our call for dimensionalizing privacy is a concurrent call for dimensionalizing digital disempowerment. At the very least, caution and reflexivity are required in discussion of these underlying concepts, as they may manifest differently in various dimensions, contexts, and situations. More fundamentally, dimensionalizing privacy invites more nuanced and deeper engagement with explanatory mechanisms underlying digital disempowerment with regard to actors in different positions within the power structure.

The challenge of studying privacy and digital disempowerment

A major source of complexity in privacy research is the definition and measurement of a concept that is intangible, ambiguous, and constantly changing (Solove, 2008; Wu et al., 2019). Established privacy theories, such as the privacy calculus or communication privacy management, are often criticized for relying on overly broad or reductive conceptualizations of privacy (Solove, 2008). Though frequently adapted for today's digital environments, such theories were largely developed prior to the bulk digitization of behaviors, identities, and social activities. They often fail to account for persistent dataveillance and predictive analytics, and consequently overlook individuals’ vulnerability to harms posed by algorithmic identification, exploitation, and discrimination (Baruh and Popescu, 2017; Lutz et al., 2020). As such, established conceptualizations of privacy may not adequately explain how individual agency is undermined in digital, computationally mediated environments.

Privacy is often theorized as an interpersonal phenomenon focused on horizontal relationships among peers (e.g. Altman, 1974; Petronio, 2002; Westin, 1967). Such conceptualizations emphasize notions of control, individual agency, or choice and are anchored in rational decision-making (Acquisti et al., 2015). A more recent shift toward appropriate flows of information as requiring contextual integrity (Nissenbaum, 2010) still places an individual actor in the center, emphasizing a socially oriented approach to privacy that relies on one's expectations, social norms, and (re)actions in interpreting privacy-protecting behavior. Such overreliance on personal autonomy and rational choice is increasingly questionable given the interconnected nature of today's communication, which emphasizes privacy as networked and not fully controlled by the individual, either socially or technologically (Dienlin and Metzger, 2016; Hargittai and Marwick, 2016). Ultimately, it also obscures the power imbalance between individuals and institutions engaged in big data analytics (Baruh and Popescu, 2017) and limits examination of the causes and consequences of digital disempowerment. At the same time, studies that explicitly engage with questions of power and the political economy of social media tend to adopt an institutionally focused perspective on power relations (e.g. McGuigan, 2023; Noble, 2018). This body of work is not designed to engage with decision-making processes made by individuals which result in digital disempowerment and withdrawal.

The way in which privacy is studied empirically (Masur, 2023) may further reify conceptual lacunas. Using established and validated measures, which are required to engage in a broader conversation across literatures, may reinforce a focus on individual-level privacy decision-making (Bartol et al., 2021); often, they envision privacy threats by a single tangible actor, either social or institutional (e.g. Yun et al., 2019). With origins in research on consumer acceptance of and engagement with e-commerce websites (e.g. Malhotra et al., 2004; Smith et al., 1996), measures of privacy concern are often adapted to study disclosure behaviors related to social media platforms (e.g. Chen and Chen, 2015; Dienlin and Metzger, 2016) where users tend to view their privacy in horizontal terms (Hargittai and Marwick, 2016; Quinn et al., 2019). This results in a violation of the compatibility principle between vertically oriented measures of privacy concern and horizontally oriented privacy-protecting behaviors.

Focusing on privacy behaviors as an outcome further complicates empirical work. Behaviors can be self-reported, but may be difficult to observe in the field, as they include “invisible” acts such as failure to connect, disengagement, or reduced levels of information disclosure. While research has established that fundamentally higher levels of privacy concerns result in greater privacy-protecting behavior (Benamati et al., 2017; Chen and Chen, 2015), many identified privacy-protecting behaviors tend to be unidimensional, such as eschewing online purchases, changing default privacy settings, or disclosing specific information (e.g. Chen and Chen, 2015). Finally, studies of e-commerce or banking typically focus on interaction between an individual and a clearly identified website, which represents the vertically oriented institutional actor, while studies of social media tend to frame privacy in horizontal terms, highlighting limitations on self-disclosure (e.g. Dienlin and Trepte, 2015) or withdrawing behaviors vis-a-vis an online community (e.g. Dienlin and Metzger, 2016). These two privacy orientations are frequently examined independently, which not only obscures the ability to understand how they may cause privacy constructs to diverge in each orientation, but also how they might overlap.

The challenge of privacy deconstruction

Capturing the multi-dimensionality of privacy and the mechanisms through which it is enacted is a difficult, but necessary step for privacy scholarship that seeks to understand digital disempowerment. Notions of contextual integrity (Nissenbaum, 2010) and context collapse (Marwick and boyd, 2014) have added complexity to privacy's conceptualization; yet disentangling privacy as it relates to individuals, institutions, and technologies remains elusive (Bazarova and Masur, 2020). Moving away from a unidimensional view of privacy is crucial for making sense of conceptually dependent ideas such as privacy fatigue, cynicism, and digital resignation. With growing context collapse in digital communication, intentional engagement in privacy management in one dimension can be interpreted as an expression of fatigue in another. For instance, in studies of privacy cynicism, which often focus on interaction between an individual and a website, privacy conceptualizations often take on a vertical orientation (e.g. Lutz et al., 2020; Segijn and van Ooijen, 2020). Such studies demonstrate that individuals, who are more concerned about their privacy are more likely to engage in privacy-protecting behaviors and less likely to engage in activities requiring disclosure of personal information. Conversely, studies focused on social media usually view privacy through a horizontal lens (e.g. Bagger, 2021; Hargittai and Marwick, 2016) and demonstrate that, despite showing privacy concerns, individuals are still willing to disclose information. While the concept of “privacy concerns” may appear comparable between these two examples, the power position of the individual with regard to the privacy-violating party is fundamentally different; similarly different is the array of privacy-protecting behaviors at the individual's disposal. In other words, one can feel competent and be proactive in managing privacy with others on a social media platform, but feel disempowered in protecting privacy vis-a-vis platform advertisers or the platform provider.

Existing research hints at the importance of distinguishing between privacy dimensions. When such dimensions are considered concurrently, Baruh et al. (2017) found that higher levels of privacy concerns are negatively associated with the use of online services (vertical orientation) yet observed no significant relationship between privacy concerns and the use of social networking sites (horizontal orientation). Lutz and Strathoff (2013) found that individuals exhibit rational, calculated considerations of privacy risks when adopting a societal, or vertically oriented, perspective. Yet, when adopting a community perspective, such as found with social media, the emotional rewards of belonging overcome rational considerations of risk. These findings, rather than highlighting a paradox, present two fundamentally different types of relationships. It is indeed possible that feelings of uncertainty and mistrust (Lutz et al., 2020) which lead to a sense of powerlessness and resignation in vertical privacy relationships do not carry over into horizontal privacy relationships. To further understand and test this intuition, scholars need explicit engagement with dimensionalized privacy-related constructs.

Disentangling digital disempowerment

To better understand processes of digital resignation and disempowerment, we also call for critical engagement with the path dependency that has been established by dominant privacy conceptualizations and measurements. While repeated use of established measures related to e.g., privacy concerns, privacy self-efficacy, privacy-protecting behaviors is beneficial for consistency and reliability, it has also constrained research possibilities. Using established measures can be akin to looking for a missing key under a lamppost merely because the light is there. In the case of privacy research, this tendency not only limits consideration of the phenomenon under study, but also carries over the potential biases and inadequacies of original measures (Arora, 2019). While some attempts have been made (e.g. Epstein and Quinn, 2020; Lutz and Newland, 2021; Masur, 2019; Masur et al., 2021), further dimensionalizing privacy measures along their vertical and horizontal orientations can address such constraints, while accounting for power differentials in computationally mediated environments. Instrumentally, this will help to avoid violations of the principle of compatibility when studying privacy-related constructs; at a more profound level, it might invite new ways of theorizing privacy, for example, by distinguishing separate theories within dimensions.

Conceptually, privacy in the singular holds increasingly limited research utility. Privacy is neither universally understood and experienced across groups and individuals, nor individually, across contexts. Those in a position of power and those who experience powerlessness often rely on different mechanisms in their aspiration to achieve the same desired outcomes.

Fundamentally, differentiating between horizontal and vertical orientations highlights the power imbalance, and the resulting sense of dis/empowerment. As such, expressions of resignation may not be equivalent to a capitulation to privacy incursion or passive acceptance of a degradation of individual agency, but rather a function of social forces. The use of invisible mechanisms of privacy protection, such as disconnection or limiting engagement on mainstream platforms which are often interpreted as resignation, can have multiple interpretations (Talvitie-Lamberg et al., 2022). For example, when one moves from Twitter to Mastodon, it can be a political act of resistance to institutional practices (vertical orientation), or a stance towards one's community of peers (horizontal orientation), or both (La Cava et al., 2023).

Empirically, deconstructing the orientation of privacy research is the first necessary step towards dimensionalizing privacy in practice. Historical emphases on vertical dimensions of privacy inadequately address horizontally oriented factors that influence digital disempowerment, such as social norms or relationships. Further, they ignore how technological contexts might influence variations on conditions such as resignation and cynicism. Efforts to develop explicitly dimensionalized measures of privacy concerns (e.g. Epstein and Quinn, 2020; Masur, 2019; Masur et al., 2021) and privacy-protecting behaviors are encouraging; however, it is important to continue developing and validating additional dimensionalized measures (e.g., privacy self-efficacy) across social and technological contexts in order to understand how privacy and its outcomes differ across meaningful units of comparison. Greater attention to varieties of expressions of resignation, particularly those that are difficult or impossible to observe (e.g., reduced levels of disclosure or non-/disconnection) will help understanding of how individuals navigate datafied environments. For example, availability of digital behavioral data, particularly data donations (van Driel et al., 2022), presents an opportunity for capturing and understanding previously unobserved forms of resignation. Studies that combine analysis of digital behaviors with regular surveys or interviews might offer further insight into the origins and varieties of resignation. Experimental manipulation of different dimensions of privacy might test for priming effects and causality. With that, we acknowledge both overly general/universal and hyper-specialized approaches to measuring privacy-related constructs are inherently limited. The former may not capture situated nuance and lean toward contextual agnosticism, while the latter inhibit comparability and the ability to generalize. Striving towards a middle ground may be preferable, to ensure that privacy measures are usable across varying and continuously changing socio-technological contexts.

Finally, while we call attention to distinctions in vertical and horizontal dimensions to privacy, this is by no means the only fruitful path. In this essay, we place the individual at the center of privacy interaction, but important and substantive information exchanges occur also among institutional actors. Such interactions implicate a similar power differential in relation to the individual, despite occurring between institutional actors, and evoke ideas of political economy and institutional power. These activities, while potentially also explaining the dynamics of digital withdrawal and resignation, invite a conceptually different perspective on individual agency, rationality, and relative position of dis/empowerment vis-a-vis institutions. Even if we focus on an individual's online experience, the relative orientation to privacy – as either vertical or horizontal – is not necessarily exclusive, distinct or unique. Other dimensions of privacy like those related to temporality may be equally meaningful. For example, privacy-dependent constructs may shift with age or over time, and temporal proximity of data may relate to its subjective sensitivity for an individual. Dimensions such as relational proximity or interconnectedness (e.g. Baruh and Popescu, 2017; Bazarova and Masur, 2020) might also be relevant to dimensionality of privacy constructs. For example, “intimate surveillance” (Leaver, 2020), such as that conducted by parents or partners, may differ in its relationship to digital disempowerment than that which is generated through corporate or government dataveillance. Other factors such as perceptions of privacy risk may prove insightful to unpacking both privacy resignation and disempowerment, as the potential effects of resignation on the privacy calculus are teased out (e.g. Meier and Krämer, 2022; Wirth et al., 2018). In addition, it is important to note that overlapping conceptualizations of privacy can occupy the same socio-technical space, as is often evident in social media where individuals simultaneously engage in both vertical and horizontal privacy relationships (Quinn et al., 2019).

In summary, we argue for a broader adoption of a dimensionalized approach to privacy as a path towards a more refined and meaningful understanding of digital disempowerment. Doing so, will advance the ways in which we think about how such disempowerment manifests in various dimensions and contexts, and in relation to disparate actors. Given the tension between the intimacy of individual privacy experiences, and the global vastness of digital platforms, we need to deconstruct privacy in order to assemble it anew.