Abstract
Context collapse blurs the boundaries between public, private, and professional selves and has emerged as an important research focus in relation to vocational identity. However, the conditions under which context collapse occur have been empirically neglected in the literature. Utilizing Davis and Jurgenson’s theoretical framework of context collapse (i.e., context collusion and context collision) and Erving Goffman’s theory of face-work, this study seeks to determine the consequential outcomes associated with the different context collapse conditions in two phases (Quasi-experimental in Phase 1 and open-ended questions in Phase 2). Specifically, a quasi-experimental study with scenarios was used to examine whether intentionality within context collusion and context collision influenced participants’ perception of loss of face and affect. First-year tertiary students (N = 151) who were also working were randomly selected from a capital state university student population and asked to respond to hypothetical online context collapse scenarios. Multivariate analysis of variance was conducted and the results indicated that context collapse has significant impact on participants’ loss of face and affect (emotion). In addition, a follow-up analysis of variance reveals partial support for the significant impact of context collapse on loss of face and affect.
Introduction
Advances in communication technologies, such as providing compelling and interactive user experiences through mobile internet devices, have fundamentally altered the way we engage and socially connect with others. According to Davis and Jurgenson (2014), individuals communicating online can adopt multiple identities which “exist within a network of others who hold particular expectations about who the actor is” (p. 477), what he or she ought to think and how he or she should behave. Importantly, the overlapping of these role identities through distinct networks can lead to collapsed contexts (boyd, 2008, 2010; Marwick & boyd, 2011; Vitak, 2012). Broadly speaking, context collapse refers to how people, information, and norms from different settings all converge into one context (boyd, 2002, 2008; Marwick & boyd, 2011; Meyrowitz, 1985). As early as 1985, Meyrowitz (1985) noted that broadcast media can speak to multiple invisible audiences who exist in different time zones, different places, and who possess different norms. Consequently, context collapse can flatten multiple audiences into a single context, making the management of the self and our online identities across varied settings increasingly complicated (boyd, 2002; Meyrowitz, 1985). While researchers have started to address the benefits and challenges associated with the meshing of social contexts online, less is known about the conditions under which context collapse occurs.
Davis and Jurgenson (2014) posit that context collapse can occur in two distinct ways, namely, context collusion and context collision. Context collusion occurs when individuals “intentionally collapse, blur, and flatten contexts,” while context collision occurs when “different social environments unintentionally and unexpectedly come crashing into each other” (Davis & Jurgenson, 2014, p. 480). Importantly, what separates context collusion and context collision is intentionality. In other words, context collusion occurs when actors intentionally bring together various contexts and related networks. For example, Facebook users who intentionally post on their Facebook account information about their current relationship status (e.g., broken off with current partner and is now single) provide insight into a situation that their followers would otherwise have no prior knowledge about. In contrast, context collisions refer to those “occasions in which contexts come together without any effort on the part of the actor, and sometimes, unbeknownst to the actor with potentially chaotic results” (Davis and Jurgenson, 2014, p. 481). For example, being tagged in photos someone else posted with no advanced knowledge this was going to occur. The key difference between the two is intentionality and it is important to note that neither has an inherent value judgment attached.
In other words, while both context collusion and context collision are examples of collapsed contexts, they are conceptually and consequentially distinct from each other. The intentionality of the individual and their role in either allowing new information to come forth—as in the case of context collusion—or if it is unbeknownst that this information will be introduced—as in the case of context collision—is crucial. We also contend that context collapse is linked to an individual’s perception of face.
As Erving Goffman (1967) argues, the term face represents the positive social value and approved social attributes that individuals assume for themselves during an interaction. Face indicates the extent to which the information presented about the self to the world is one that remains internally consistent and is supported by judgments and evidence of others occurring in a specific situation (Goffman, 1967, p. 6). Face is therefore not physiognomic but rather is “diffusely located in the flow of events in the encounter and becomes manifest only when these events are read and interpreted for the appraisals expressed in them” (Goffman, 1967, p. 7).
Social media provides a setting for the public to “create, gather, discuss, engage, debate, and share information; representing contexts for individual self-expression and collective association in which users who use these technologies possess relatively little influence on their architecture and governance” (West, 2018, pp. 4366–4367). A study by the Pew Research Centre (Greenwood et al., 2016) found that about 56% of adult social media users in the United States used more than one of the following platforms: Twitter, Instagram, Pinterest, LinkedIn, and Facebook. The ability to display oneself and to engage in various social contexts online has provided a way for individuals to enhance their identity, to obtain resources and/or to take advantage of all the affordances of social media. However, there are also times when social media use can potentially work against an individual’s perceived sense of self (Davis & Jurgenson, 2014). For example, when an actor’s multiple online identities blur into one another leading to confusion about an individual’s sense of self (boyd, 2008, 2010; Marwick & boyd, 2011; Walsh & Baker, 2017).
Literature Review
Goffman (1967) defines the term face as the positive social value effectively claimed by a person by the line they assume for themselves during an interaction. In other words, face can be described not as the literal visage of the person but the metaphorical “image of self-delineated in terms of approved social attributes” (Goffman, 1967, p. 5). A person might be said to be in face when the line they claim for themselves is one that is “internally consistent, that is supported by judgments and evidence conveyed by other participants, and that is confirmed by evidence conveyed through impersonal agencies in the situation” (Goffman, 1967, p. 6). Face therefore is not the property of the individual, but collectively bestowed during interaction. As Goffman articulates, face “is something that is not lodged in or on his body, but rather something that is diffusely located in the flow of events” (Goffman, 1967, p. 7).
Managing Face During Context Collapse
Context collapse is critical because individuals can be seen to either “lose face” or be in the “wrong face” for the situation that they are located. Goffman (1967) contends that when we lose face, what occurs is the introduction of information that “cannot be integrated, even with effort, into the line that is being sustained” (p. 8) such as when an individual is subjected to a joke at his or her own expense. This represents an interaction where the face of the other is targeted; the individual may lose or may show a wrong face because such joking typically requires insider knowledge that is withheld from the target of the joke. The comedic element therefore emerges from the situation in which the individual has been placed (a situational naivety with respect to the joke). The individual therefore responds without levity and the gap in understanding how this situation is defined (as a humorous conversation) renders the interaction potentially comical to observers who are witness to the face that is erroneously introduced. It is this tension in interaction that suggests why jokes can be risky; there is uncertainty as to if the target of the joke will comply graciously with being caught in the wrong face or if they will take umbrage at being deliberately placed in a situation where they might lose face, compounding the awkwardness of the interaction.
In managing face, individuals shape how others view their face through verbal and nonverbal behaviors and online, these aspects of interaction become mediated to platformed “exhibitions of the self” (Hogan, 2010). When we maintain our face, this means the line taken in our performance remains consistent and is “supported by judgements and evidence conveyed by other participants” (Goffman, 1967, pp. 6–7). However, when we lose face, this projected presentation of the self, falters. Here the performance that we are attempting to foster is damaged or undermined either by an interlocutor or our own inabilities to maintain the prevailing performance one is attempting to exhibit. As Smith (2006) notes, sometimes “discrediting information crops up, leading the person to be ‘in the wrong face.’ Sometimes the person is ‘out of face’ i.e., unready for the encounter” (p. 51).
In situations when context collapse occurs, face will potentially become threatened given the version of the self that one presents may be contradicted or undermined contingent on the form of context collapse. Importantly also, the distinction of context collapse and its variants of context collusion and context collision are reminiscent of Goffman’s formulation of how communication can be categorized into expressions that are “given” and those that are “given off” (Goffman, 1959, p. 2). As Smith (2006) notes, the expressions that we give are “taken to be intended; those we give off are often assumed to be unintentional” (p. 35). Although we tend to focus on the narrower forms of communication that are “given,” it is the ones that are “given off” that are the more “theatrical and contextual kind, the non-verbal, presumably unintentional kind, whether this communication be purposely engineered or not” (Goffman, 1959, p. 4).
During context collapse where expressions “given off” occur, and especially in terms of context collisions, occasions are brought together unbeknownst to the actor placing them in new and potentially risky ambiguous informational contexts (Duguay, 2016, p. 898). This may also impact certain affects within individuals. There may be times in our interactions with others when our face is challenged or undermined in some ways which then invoked certain affects (Goffman, 1955). Affect here refers to the underlying experience of feeling, emotion or mood (Hogg et al., 2010). For instance, suppose when after telling a funny story or joke, one friend states that, “You know, you are trying way too hard. You are really not funny.” Your friend’s comment challenges your image (face) as a funny person and this can lead to feelings of hurt, anger and/or embarrassment.
While jokes tend to exploit inconsistencies in face, there are more threatening situations where a loss of face can be demonstrative of a significant and lasting reputational problem. When a person is in face, they are said to “respond with feelings of confidence and assurance” but when out of face, this means the events that are part of a situation “cannot be readily woven into the expressive fabric of the occasion” (Goffman, 1967, p. 9). A mundane example of this process is when an individual places a considerable bet on a racehorse and boasts to their friends the horse is a “sure thing,” only then for that horse to come in last during the race. A loss of face therefore might be registered through emotions of embarrassment, shame and inferiority due to the potential longer-term implications for one’s reputation. Furthermore, the individual may “feel bad” that they relied on a projection of themselves which now is seen as erroneous and therefore threatens their reputation. The emotions Goffman (1967) identifies as part of how an individual feels when inadvertently situated “out of face” are revealing: he suggests they may be “taken aback,” “confused,” feel “momentarily incapacitated,” “falter,” “collapse,” “crumble,” become “embarrassed” and “chagrined” (p. 8). Ultimately, the individual is likely to become shamefaced and experience emotions that stem from being unable to integrate situationally apt information, nor perceive oneself as appropriately situated in a context in which they were originally accustomed.
Impression Management and Social Media
Face-work is the protective process by which individuals respond to and compensate for information that is introduced that could potentially lead to an irreconcilable version of face. The function of face-work therefore assists to manage interactions, but this becomes complex in the context of the increasing use of information technologies (Lim et al., 2012; Walsh, in press; Walsh & Baker, 2021). Researchers have found that users communicate emotions and personal experiences on social media to represent themselves and to also seek social support (Wang et al., 2016). For example, Kim and Lee (2011) found that many users tend to highlight socially desirable parts of their identities and adopt a “positive self-presentation” which was found to increase subjective well-being. Others such as Lee-Won et al. (2014) found that sharing negative and positive emotional experiences have benefits for users such as having a positive public self-consciousness. Public self-consciousness here means users being aware of how other individuals might perceive their appearance or their action in the public domain. The question, however, is under what conditions users express positive versus negative representations. One assumption could be that users may be more affected (in terms of their face and emotions) under context collusion compared with context collision because knowingly doing something inappropriate or wrong is less socially acceptable.
Managing the Self at Work
Our sense of self refers to our perceptions of what defines us as an individual. This can include our personality traits, our likes and dislikes, our beliefs about certain topics and how we interact in different situations. As Nippert-Eng (1996b) suggests “the self becomes separated, parcelled out so that certain aspects of identity are emphasized in one realm, others in its opposite” (p. 34). Consequently, individuals aim to present different versions of the self specifically in terms of projecting personas that remain professional and consistent at work, which can become acute when individuals experience context collapse.
The process of managing the self at work has shifted in recent times where recreational life and work activity has become more integrated and connected across different times and spaces (Thomas et al., 2017) especially given the rise in mobile work (Cohen, 2010) which has progressively decoupled work activity from the physical confines of the office or workspace (Hill et al., 2003). To work effectively, employees are likely to learn practices to manage the boundaries and continuities between not only work and home activities but also versions of the self that are shaped depending on where a person is situated and the activities in which they are engaged in (Loh et al., 2010; Nippert-Eng, 1996a, 1996b).
In addition, the nature of work activity has also transformed with much work requiring the incorporation of emotional labor (Kim & Hollensbe, 2018). As Hochschild notes with respect to emotional labor across various professions, many variants of professional life can reveal the importance of managing emotions at work (Hochschild, 1983, p. 147). The practice of emotional labor is transposed to our online working lives given much work now is actively mediated beyond co-present situations to working in digital contexts and across social media platforms (Bridgen, 2011; Evers, 2019). Face and the affect associated with its management by the individual are important because they not only represent potential reactions to others but also provide an indication of the ways we interact with our environments (Clark et al., 2004), that increasingly include digital contexts that potentially expose individuals to context collapse.
While originally the use of computers was characterized by some scholars as a context for the creation of a second self that was associated with a level of freedom that enabled the expression of feelings and emotions (Turkle, 1984), the contemporary use of social media has now transformed into a crucial dimension of the identity of workers that now requires careful management (Abril et al., 2012; Ollier-Malaterre & Rothbard, 2015). Employees are increasingly aware of the role their social media footprints can play in obtaining and maintaining employment (McDonald et al., 2016). Critical here is the situation where social media has the potential to lead to context collapse in work contexts. Given the proliferation of communication technologies and the corresponding increase in the potential for context collapse, this renders work as a site where impression management and face-work come into play.
Several media and communication scholars have examined the idea of context collapse and explore how different versions of media function to enable this collapse in different ways (boyd, 2008, 2010; Marwick & boyd, 2011; Walsh & Baker, 2017). In this study, we experiment with the way this collapse occurs with respect to context collusion where media users are aware of the collapsing of different situations, as compared with those who have no agency or forewarning that this collapse is to occur (i.e., context collision).
The theoretical antecedents of context collapse can arguably be found in the work of Meyrowitz’s (1985) and his extension of Erving Goffman’s (1959, p. 208) idea of impression management and the role that social situations play in ensuring our interactions and behaviors remain socially responsive to the context the individual finds themselves located. As Goffman (1983, p. 2) suggests, it is the human condition that for most of us, our daily life is undertaken in “the immediate presence of others; in other words, that whatever they are, our doings are likely to be, in the narrow sense, socially situated.” For example, the language we use, and behaviors adopted as parents in negotiating with a child at home to complete a given chore will situationally require distinct approaches when undertaking a similar negotiation with a colleague at work.
In the context where we are increasingly tethered to communication technologies where we adopt “always-on and constantly connected selves” (Turkle, 2008), the role of managing interaction becomes more complex given the greater exposure to different situations and people, but also arguably increases the importance of face-work in assisting these interactions. As Davis and Jurgenson (2014) argue, it is inaccurate to suggest that context collapse is absent from face-to-face situations or that it simply emerged with social media or Web 2.0 technologies (p. 479). But with social media, “some contexts become more porous and allow outside networks and roles to easily seep in whereby users actively and intentionally invite various contexts” (Davis and Jurgenson, 2014, p. 481). Therefore, our study will investigate and explore how workers that simultaneously study, undertake identity-work and how they manage their professional personas. We also consider how these individuals emotionally respond to the possible tarnishing of their professional persona such as when their embarrassing photos are uploaded onto social media.
Moral Emotions
According to the framework of moral emotions, moral emotions provide the motivational force—the power and energy to do good, and to avoid doing bad (Kroll & Egan, 2004). Haidt (2003) defines moral emotions as those “that are linked to the interests or welfare either of society as a whole or at least of persons other than a judge or agent” (p. 276). Shame, guilt, embarrassment, and pride function as conscious elements of the emotional moral barometer that provides individuals with what is socially and morally acceptable in society and are directly linked with Goffman’s formulation of face-work (Scheff, 2003). Therefore, individuals who undertake inappropriate action, such as where participants agree to have an embarrassing photo uploaded online if they lose a bet will experience more negative affect. This is because intentionally uploading someone’s embarrassing photo onto social media does not constitute a civil or socially acceptable thing to do. This intentionality means that individuals understand the likely consequences associated with inappropriate thoughts and/or actions. Thus, affected individuals rightly assumed they should experience negative affect. In contrast, in occasions where contexts come together without any intent from the actors such as when an actor was drunk and they were then tagged in photos someone else posted (e.g., context collision), the consequential reaction will be less severe compared with the previous case (e.g., context collusion).
Aims and Objectives
One problem that arises from the increasing prevalence of social media use is context collapse where an actor’s identity, behaviors, and their associated environments can blur or seep into one another (boyd, 2008, 2010; Marwick & boyd, 2011; Walsh & Baker, 2017). While academic literature is beginning to investigate context collapse, less is known about the conditions under which context collapse occur or the consequential impact of these experiences on an individual’s sense of self (Davis & Jurgenson, 2014). In an effort to further understand the impacts of context collapse, we draw on sources referenced so far to conduct an empirical study to examine the conditions under which context collusion and context collision occur and the impact this can have on an individual’s affect and face. Therefore, the aim of our study is to investigate whether context collusion and context collision can affect an individual’s face and emotions (i.e., positive and negative affect).
To help us achieve this aim, a quasi-experimental research design (Phase 1) using three randomly assigned scenario-based hypothetical situations were randomly distributed to participants to test the conditional differences within context collapse. Specifically, participants were asked to imagine themselves in situations in which their social media usage led to (1) context collusion or (2) context collision. Participants were then asked to respond to a series of questions that measured (1) their sense of face and (2) affect. A control group was also tested without context collapse. A qualitative study (Phase 2) was then conducted as a follow-up to understand the complexity and meaning associated with social media use, the strategies (if any) they adopted in managing their social media, and the consequences associated specifically with context collusion and context collision. Therefore, we predicted the following:
H1: Context collapse has a significant impact on participants’ loss of face and affect (emotion).
H2: Participants who encounter context collapse (in the form of context collusion) will experience more negative affect, more loss of face and less positive affect compared with participants who encounter context collapse (in the form of context collision) or to participants who encounter no context collapse (i.e., in the control group).
To test the above hypotheses, a quasi-experiment using three randomly assigned scenarios were utilized and participants were asked to imagine themselves as follows. In Scenario A (experimental group A: context collision), participants encounter a scenario where they were under the influence of alcohol and found out that someone has taken an embarrassing photo of them and uploaded it onto social media. In Scenario B (experimental group B: context collusion), participants encountered a scenario where they lost a bet and agreed to have their embarrassing photo uploaded onto social media. Participants in Scenario C (control group: no context collision or context collusion) encountered a social context scenario without any context collapse (see also Appendix A). Please refer to the Appendix for more information on each of these scenarios. A detailed discussion of these scenarios is also provided in the materials section. Participants were then required to respond to a series of questions that measures their sense of: (1) loss of face and (2) affect (i.e., negative and positive affect). Information about the measures themselves is provided in the “Materials” section.
Method
Participants
A total of 151 undergraduate Australian university students (55 males, 96 females; mean age = 22.94 years, SD = 6.27) in a capital state university participated in this quasi-experimental study. To be eligible for the study, students had to be 18 years and above and must be currently employed. In selecting our sample of working students, we also aimed to recruit participants across a diverse range of subjects from different university disciplines such as Accounting, Law, Business Management, Nursing, Psychology, Policy, Sociology, and so on. Consequently, these working students were found to have vast different life experiences and were employed in the service industry (63.4%), followed by professional consultancy (6.2%), trade (9.3%), education (5.6%), clerical (5.6%), community service (4.3%), health care (3.7%), financial services (1.2%), and construction (0.6%).
Procedure and Measures
Once ethics approval was granted by the University’s Human Research Ethics Committee (Project ID XXXX, University of Canberra), unit conveners were contacted via emails for their permissions to recruit students. The lead author then attended one of the lectures of the unit convener and recruited students during the last 30 min of the lecture. Students were provided with a cover sheet that explains the aim and purpose of the study. Students were also told that participation was voluntary, that the study was confidential, and that they can withdraw without any penalty during the experiment. Students interested in the experiment were instructed to sign the consent form and were then randomly assigned into (1) context collision, (2) context collusion, or (3) control groups. Interested participants’ names were then entered into an excel worksheet and the formula RAND() was entered into the formula bar so that participants can be randomly assigned into one the three groups. All students who have consented to the study were instructed to read their respective scenarios carefully and to image themselves in the situation described before responding to the scenario questions.
Analysis
This quasi-experimental study employed a one-way multivariate analysis of variance (MANOVA) to test the hypothesis that there would be one or more mean differences between conditions of context collapse (context collusion versus context collision), loss of face and affect. Students were presented with randomly distributed vignettes describing hypothetical social media with context collusion or context collision or control group with no context collapse. Students were required to read their respective scenarios carefully and then to respond to a series of questions that measures three dependent variables (1) loss of face, negative affect, and positive affect. To reduce biases and to ensure good validity, social desirability and manipulation measures were used as part of the MANOVA analysis. Finally, participants were asked to write a paragraph to describe their feelings about the vignette and what strategies (if any) they used to manage their social media profile. These responses were then thematically collated as they emerged from the data analysis.
Materials
The vignettes were created specifically for the current research and were developed in accordance with recommendations by Gould (1996). A vignette is a carefully written brief description of either a person or situation designed to simulate core characteristics of a real-world scenario (Atzmüller & Steiner, 2010; Gould, 1996). To ensure the effectiveness and realism of the vignettes, researchers generally construct their vignettes based on existing literature and case studies (Atzmüller & Steiner, 2010; Gould, 1996). Therefore, the vignettes created in the current study were patterned from investigation regarding how participants understand their social media use (Beninger et al., 2014).
Scenarios/Vignettes
Scenario A and Scenario B first asked participants to imagine themselves in a scenario. The scenario depicts a situation where participants used social media to keep up to date with their friends and family. Participants were told within the scenarios that they also used professional employment-oriented online services such as LinkedIn to maintain and to stay connected to their professional network. Participants were informed that their online profile contained their names, job title, and the company’s name (participants insert these themselves). Participants were then asked to imagine that they have posted online about their social, work life, and work projects they are currently working on to develop a strong professional online image. However, when participants logged into their social media account a few days back, they found a photo of themselves in a compromised state (e.g., state of drunkenness at the annual office party) uploaded by a friend onto social media. In Scenario A (experimental group A: context collision), participants while under the influence of alcohol had their photo taken and were tagged in their friend’s photo. In Scenario B (experimental group B: context collusion), participants were told that they have made a bet with their friend and the penalty for losing was to have their photo posted online. Scenario C (control group) depicts a friend having posted a “normal” photo of participants and their colleagues at the annual end of year office party. Please refer to Appendix A for full details of the scenarios. Following the scenarios, participants were asked to respond to questions that evaluated the following dependent variables: loss of face, positive and negative affect. Please refer to Appendix B for full details of all items used.
Loss of Face
Loss of face was measured using Zane’s (2000) 21-item Loss of Face (LOF) measure. This measure uses a 7-point Likert-type scale with responses ranging from 1 = Strongly disagree to 7 = Strongly agree. An example item includes, “I am more affected when someone criticizes me in public than when someone criticizes me in private.” There was no reverse scored item. The reliability for the LOF in the current study is .87 demonstrating good internal consistency.
Affect
Affect was measured using Watson et al. (1999)’s Positive and Negative Affect Scale (PANAS). Participants were asked to describe their emotions after experiencing the incident using a scale ranging from 1 = very slightly or not at all to 5 = Extremely. Watson et al. (1988) reported reliability for the positive affect to be between .86 to .90 and the negative affect scale to range from .84 to .87. The Cronbach’s alpha score for positive affect was .79 and for negative affect was .93.
Checks
To ensure that the contents of the vignettes could be accurately recalled and that participants correctly stated their knowledge of context collapse, manipulation checks were administered after each vignette. In Scenario A, participants answered yes or no to the question about their knowledge that someone has uploaded a photo of themself. Similarly, in Scenario B, participants answered yes or no to the question about their knowledge that someone has uploaded a photo of themself.
To address common method variance, participants were advised that there were no right or wrong answers and to therefore answer the questions as honestly as possible. In addition, open-ended questions posted at the end of each vignette provided participants the chance to freely express their opinions and to gauge what they might have done in the situation. For example, participants were asked to write a paragraph to further describe their feelings about the vignette. As there is a potential for participants to report socially desirable responses, Stöber’s (2001) Social Desirability Scale (SDS-17) was administered at the end of the vignettes. Higher scores on the SDS-17 indicated a stronger tendency to present oneself in a socially desirable way (Stöber, 2001).
In the present study, the manipulation checks showed that 98% of participants correctly identify the conditions between the groups. Social desirability scores from the SDS-17 ranged from 1 to 15 (M = 9.31, SD = 2.92) which suggested that there were no strong social biases among participants.
Statistical Analysis
A one-way multivariate analysis of variance (MANOVA) was conducted to test the hypothesis that context collapse significantly affected participants’ outcomes (face and affect). This was then followed up with univariate analyses of variance (ANOVA) to examine if participants in context collusion would be more likely to experienced heighted negative affect and loss of face compared with participants in context collision as well as participants in the control group. The ANOVA was adjusted at the Bonferroni alpha level of .025 to determine whether the overall effects were significant. All statistical analyses in the current study were conducted using SPSS version 25.
Next, we compared the themes from the open-ended questions with the statistical findings to determine whether they shared any similarities or differences across the three groups of participants (i.e., experimental group A, experimental group B and control group). We use thematic analysis to identify, analyze, and report patterns (themes) within data (Braun & Clarke, 2006). We read, re-read, and analyzed participants’ responses until saturation was reached, that is until no new information was gathered from the organization of the themes that had been developed out of the analytical process. These themes that we produced from the analysis were first identified, coded, and tabled across the three groups of participants. Themes and sub-themes were isolated. Next, emerging common themes and sub-themes from individual groups were analyzed before a group analysis was undertaken across the three groups.
Results
The correlation matrix, means, and standard deviations for the three dependent variables are presented in Table 1. Pearson’s correlation test was conducted to identify the relationships between the dependent variables, and correlation coefficients showed small to moderate correlations between the dependent variables. Mahalanobis’ Distance (MD) was conducted to identify potential outliers. The critical value for 3 dependent variables was 16.27 and an outlier with Mahalanobis’ Distance value of 17.55 (which exceeded the critical value of 16.27) was excluded from the final analysis. In other words, if the Mahalanobis’ Distance value exceeded 16.27, we excluded these data from the data set. In our study, we excluded the data with Mahalanobis’ Distance value of 17.55. Next, the multivariate normality test retuned a non-normality for positive affect (p < .05) and as a result, a variable transformation using Lg10 was conducted for positive affect. This transformed variable for positive affect was used in the final MANOVA analysis as the transformed data are now normally distributed. Finally, the homogeneity of variance–covariance matrices, linearity, and multicollinearity were verified and all met the assumptions for MANOVA. As not all assumptions for MANOVA were met (e.g., non-normality), the Pillai’s Trace test was applied. In addition, after a significant difference was found through MANOVA, a Bonferroni correction was made to check the type I error, and the data were tested at a significance level of .025.
Correlation Matrix and Descriptive Statistics for Dependent Variables.
SD: standard deviation.
p < .05, **p < .01, ***p < .001.
Means and standard deviations for the dependent variables in each vignette (see Table 2) revealed that participants experienced greater loss of face during context collusion, suggesting that participants who behaved inappropriately (e.g., making a risky bet) felt more embarrassed than participants in context collision. However, when it comes to affect, participants who experienced context collusion felt less negative affect and less positive affect compared with participants who experienced context collision.
Means and Standard Deviations for Dependent Variables by Scenarios.
SD: standard deviation.
Main Effect—Hypothesis 1
In Hypothesis 1, we predicted that a main effect exists between types of context collapse conditions on loss of face and affect. The main effect was found to be significant and that there were significant differences between the three conditions of context collapse (i.e., experimental group A, experimental group B and the control group) on both loss of face and affect. In other words, there was a statistically significant difference in the three types of context collapse conditions on participants’ loss of face and affect, F(3, 146) = 6.23, p < .000; Pillai’s Trace = 0.22, partial η2 = .16. The result indicates that the three types of context collapse conditions have a statistically significant effect on loss of face, F(2, 148) = 3.83; p < .01; partial η2 = .05, negative affect, F(2, 148) = 13.09; p < .000; partial η2 = .15, and positive affect, F(2, 148) = 9.71; p < .000; partial η2 = .12. Alpha correction using Bonferroni correction was used and adjusted to p < .025.
Multiple Comparisons Across Context Collapse Conditions
Loss of Face
Given the statistical significance of these univariate main effects, post hoc test via Scheffe was then used for multiple comparisons. The findings showed that the mean scores for loss of face were statistically significantly different between the control group and participants in experimental group B (p < .000), but not between participants in experimental group A and experimental group B (p = .123) or for participants in experimental group A and the control group (p = .780).
Affect (Negative and Positive Affect)
Moreover, the findings for the mean score on negative affect show that there was a statistical significance between participants in experimental group A and the control group (p < .000), between participants in experimental group B and the control group (p < .001), and between participants in the control group and in the experimental group A (p < .000) and between participants in the control group and participants in the experimental group B (p < . 001). There was no statistical difference between participants in experimental group A and experimental group B for negative affect. Participants in both these experimental groups were negatively affected. Finally, the results showed that the mean scores for positive affect were statistically significant between participants in experimental group A and the control group (p < .000), but not between participants in the control group and in experimental group B (p = .130), or between participants in experimental group A and experimental group B (p = .040).
Therefore, the overall results for Hypothesis 2 were partially supported. In other words, participants’ loss of face and affect were not significantly impacted by different context collapse conditions. However, participants in experimental group A and experimental group B responded differently to respondents in the control group for the dependent variables, loss of face and negative affect.
Participants’ Experiences Across Types of Context Collapse Conditions
Using thematic analysis (Braun & Clarke, 2006), we derived themes and patterns of experiences from the three groups of participants regarding their loss of face and affect in different scenarios (i.e., experimental group A, experimental group B and control group). Similarities and differences in respondents’ experiences were coded. Next, patterns with particular reference to the loss of face and affect were compared across the three groups of participants. Overall, results from this thematic procedure suggest participants in both experimental group A: context collision and experimental group B: context collusion tend to be more affected than participants in the control group.
According to Goffman (1967), when face becomes threatened and no longer accords with the participant’s feelings in a given situation due to contradictory information either about the individual or because of the unfolding events, individuals may experience confusion, surprise, helplessness, embarrassment or distress (Goffman 1967, p. 8). This is because the person senses less control over the social information present in the situation and this in turn results in a contradiction, in terms of the “line” they sought to establish as a professional. This is consistent with the findings in our study. For example, one respondent (from experimental group A: context collision) expressed a sense of helplessness due to her lack of control over the situation, “Helplessness, as the photo was posted against my will or consent. . .” Two respondents, one (from experimental group A: context collision) and the other (from experimental group B: context collusion) commented about their emotions when someone introduced private information about them publicly: Depending on the severity and repercussions, I would feel great annoyance and frustration. I am quick to forgive and forget but I would still be angered and betrayed by whomever posted the photo of me. I’d then feel nervous and anxious about how it has impacted my work life. Even though I was sober when I made the bet with my friend, I am angry that they would still post an embarrassing and career endangering photo of me.
The open-ended responses also suggest that when losing face, participants will judge the extremity or the significance of this loss not merely by the fact that they have lost face, but rather they assess it by what the loss signifies to them. Indeed, one respondent (from experimental group A: context collision) stated that the collapse made him angry because as he reported, “Anger, I have a professional online profile and these photos jeopardise that image and therefore my career.” Similarly, a respondent from (experimental group B: context collusion) stated that he was, “Hurt that my friend would put my professional appearance at risk when I was clearly drunk and angry at myself for agreeing to it . . .”
In general, the emotions experienced by participants in this scenario suggest that feelings were not only directed inwardly at their own conundrum, but also outwardly to account for the actions of others who may have placed the individual in the situation where face was lost. This is because as Goffman (1967, p.27) indicates, “one’s face, then, is a sacred thing” and as a result face-work will be undertaken either by the individual or those who witness the loss of face with compensatory efforts to restore face. In usual circumstances, the explicit undermining of face is less common and therefore retribution or at the very least, an understanding for why such a predicament has occurred will be sought. For example, two participants (from experimental group A: context collision) and one participant (from experimental group B: context collusion) group indicated that they . . .would have felt like a disappointment, extremely embarrassed, very distressed, highly emotional and angry, grief and angry with the person who did it. . . .would be concerned the photo may impact my future and paint an image which does not reflect my true nature. I would be very disappointed in that person. . . . would have felt embarrassed and worried that people from work could see it. Also angry at my friend for posting it as she knew about my work profile and that is morally wrong to do to a friend.
Implied in these statements is a multifaceted emotional response to the faltering of face-work, one that is registered as an internal response and against the responsible individual’s actions.
Another respondent reported, I would have felt very anxious and overwhelmed. My brain would be running possible scenarios and their consequences. I would also feel very irritated as well as we are all not children anymore.
In this respect, the undermining of a person’s face is deemed as something akin to less mature behavior. Such behavior can also be perceived as something understood by some participants as a form of intimidation: I’d feel mostly upset because that friend might think it was funny but harassing someone could never be fun. And he or she probably would have known my efforts in making a professional image for myself.
And finally, another participant (from context collision) reported feeling afraid for one’s reputation and a sense of shame that has emerged from the collision of contexts: I am scared as I would not want to ruin my reputation and ashamed as I was just trying to have fun but then it got out of hand. Should have known better.
Another participant (from context collusion) acknowledged that he was not blameless for the situation. Nevertheless, he still believed that his friend should have checked with him first before posting the embarrassing photo: I would have been upset but would be aware that I was not blameless for the situation. However, I would think it would be better if my friend had not put the photo up or checked with me.
In contrast, respondents in the control group were less emotionally upset compared with respondents in both context collision and context collusion scenarios. The majority of respondents in the control group reported that it would depend on the type of photo that was uploaded. For example, the following excerpts demonstrate this concern: Depending on the photo. If it is revealing and unprofessional then I would tell them to take it down. Otherwise, if it is something okay, I wouldn’t mind. It depends on the actual photo. If the photo could compromise me in any way, then I would be angry and feel betrayed. Otherwise, it wouldn’t really bother me. I think it would depend on what type of photo they uploaded. If I looked good in the photo, I will be happy but if I looked bad, I’d get annoyed.
Participant Strategies in Responding to Context Collapse
Respondents adopted several strategies to manage their online professional accounts, suggesting a relatively high use of privacy setting among these individuals (Litt, 2013). Some of these range from sequestering professional and friendship groups and as one participant suggests, by using “high privacy settings and only accepting certain friend’s request.” In this case, participants use the systems and affordance of social media to maintain a semblance of control over particular types of information. For example, I stay very private within my social media. I keep work and outside life distinct whilst not posting or communicating with others at all. I only use social media for news and contact with a few close friends.
Similarly, another participant indicated that he Ensure that I have my privacy settings on private so only my friends can see my photos and I need to approve photos I am tagged in before they appear on my profile.
And: I try to keep all my accounts private as well as not uploading any compromising images or information about myself.
These practices indicate that the face-work conducted by working students regarding their online selves suggest a careful management of social media and one that in some cases eschews posting freely about their personal and work experiences. As one participant suggests, “I prefer not to post online for professional or personal reasons. I minimise my persona online as much as possible.” This is indicative of a wider theme that was detected from the open-ended responses provided by participants who experienced context collision. Participants in this sense were very careful about what they posted online to ensure that their image of face is maintained appropriately when communicating with others online: I try not to post anything that would ruin my professional image and be careful what I do around people. . ..
And: I post only positive images to reflect that I live a well, social and stable life currently.
A strategy of curtailing context collapse on social media is therefore articulated and one that seeks a cautious, deliberate, and tentative approach. For example, one participant suggested that even though she was . . . active on social media (Snapchat and Instagram only), by active, I mean liking other peoples’ posts etc. However, I don’t post very often, and I am very conscious about who can see my profile and activity.
Another participant suggested, I try not to post much about work so that people see me as myself and not my work persona. I don’t post embarrassing photos in case people from work see my profile.
And similarly, I always check that nothing I say can be taken incorrectly in order to maintain my professional persona. I also have privacy settings.
Careful management of face-work activity on social media is undertaken in a relatively sophisticated manner and in a way that ensures most risk and exposure that might compromise the individual’s performance is minimized.
Discussion
Context collapse and the misalignment of imagined and actual audiences can have serious, often negative, consequences for participants that use social media. These consequences can be particularly relevant to workers that simultaneously study in that they are answerable not only to themselves but also their employers. For instance, employees can face embarrassing or career-threatening repercussions over their social media posts (e.g., Grusin, 2015; Herzog, 2015). Problematic situations can stem from the merging of personal and professional audiences when context collapse occurs. While the meshing of social contexts online has been examined, there has been limited research on the conditions under which context collapse occurs. The current research addressed this gap by drawing on Davis and Jurgenson’s (2014) theoretical framework of context collapse (i.e., context collusion and context collision) and Erving Goffman’s (1967) theory of face-work to understand the impact of context collapse and its condition on face and affect.
Overall, the results are in line with the proposed hypotheses. Hypothesis 1 was fully supported and was consistent with both Davis and Jurgenson’s (2014) theoretical framework of context collapse (i.e., context collusion and context collision) and Goffman’s (1967) theory of face-work where context collapsed (Levine & Crowther, 2008). The result showed that context collapse has a statistically significant effect on loss of face and affect (i.e., negative affect and positive affect). In other words, our sample of working students in both context collision and context collusion conditions were more emotionally upset and suffered greater loss of face compared with participants in the control group. This finding was also evidenced in many of the qualitative responses from our sample of working students, when comparing those who experienced context collapse and the control group. Specifically, participants registered multifaceted emotional responses ranging from anger, distress, embarrassment, a sense of helplessness and anxiety when context collapse occurs outside of their control. Interestingly, many working students in Scenario B (experimental group B: context collusion) acknowledged that even if they were partially responsible for the collapse (i.e., they lost a bet and a photo was posted online), they felt that their friends and colleagues should have known better than to post their photo online and rationalized this, based on the interlocuters’ ethics and morality. This line of reasoning is consistent with the framework of moral emotions (Haidt, 2003; Kroll & Egan, 2004) where victims feel righteous emotional outrage when they have been treated in a socially unacceptable manner by a perpetrator. This then may lead participants to assign blame external to themselves (Jones et al., 1972; Weiner, 1986).
We only found partial support for Hypothesis 2. In other words, participants in Scenario B (experimental group B: context collusion) suffered statistically more loss of face compared with participants in the control group, but this effect was not statistically significant between participants in experimental group A: context collision and in experimental group B: context collusion or for participants in experimental group A: context collision and the control group. In terms of affect (negative and positive affect), there was no statistically significant difference reported among participants in experimental group A, experimental group B, or the control group for positive affect. However, for negative affect, participants in experimental group A: context collision and experimental group B: context collusion were more affected that participants in the control group. A plausible explanation for this may be that context collapse is a dramatic event for all participants irrespective of the types of conditions. When context collapse occurs, this collapse can represent an attack toward one’s self and a victim in this situation is likely to experience emotional upheaval associated with this situation.
An analysis of the qualitative responses from the participants adhere to this evaluation that context collapse is in itself a dramatic event. Apart from those in the control group, participants from both contexts (i.e., context collision and context collusion) seemed to be equally affected by the collapse for both loss of face and affect. As a result of this, participants managed their faces and work activity carefully to ensure their image of self is not polluted or defiled. To do this, our sample of working students adopt several online management strategies such as (1) engaging privacy settings, (2) dutifully censoring what they post online while carefully curating who they accept as online friends, and (3) separating personal information from work information online where possible.
Limitations
The main limitation of this study was the use of a student sample which has been argued to be too homogeneous and thus restricts the ability to generalize the findings of the study outside the student population. However, we believe that the use of the student sample is justified in this study as there is good fit between the research question and the method we adopted. Moreover, the issue of context collapse is a current social issue faced by many young people given their marked use of social media. In this study, we took steps to ensure that a diverse student sample was collected across different disciplines. In addition, one of the selection criteria for this study was that participants had to be currently working. This is because we believe that working students tend to have more diverse life experiences compared with non-working students and more experience in managing context collapse given the need to transition between work and study. Another limitation was that that the research design may have failed to fully differentiate between the emotions caused by two different contexts such as emotions about self-image, versus, emotions experienced when betrayed by a close friend. Future researchers could structure the context collapse information posted for one audience to be accidentally discovered by another audience from the workplace. For example, in their study of university student bloggers, Tian and Menchik (2016) found that bloggers’ perceptions, strategies, and disclosures evolved over time in response to different audiences. Therefore, future researchers could develop a vignette that investigates how participants would feel if the information they posted for one group (e.g., close friends) was discovered by another group (e.g., colleagues or supervisors). Such a study will help differentiate between the emotions caused by different situations. Finally, this study was conducted in a capital city where internet access is readily available. Future studies could consider researching in more remote regions where internet access may be limited or perhaps in smaller rural communities where participants are less likely to experience anonymity in the community.
Despite these limitations, a strength of this study was the use of a quasi-experimental research design which extend context collapse theories into the real world. It is recommended that future research should build on the current findings by investigating how individual factors or cultural factors may influence participant’s loss of face and affect.
Practical and Theoretical Implications
The present study revealed that context collapse is a highly emotive and concerning issue for individuals, especially among working students who are seeking to foster and build their professional image. This study was able to provide an integrated and holistic explanation using Davis and Jurgenson’s (2014) theoretical framework of context collapse (i.e., context collusion and context collision) and Goffman’s (1967) theory of face-work to explain the impact context collapse has on loss of face and affect.
The implications from the current study’s finding suggest that there may be several practical solutions that workers can consider adopting to manage their face. For example, participants could actively use privacy settings embedded in social networking sites to ensure any content that tags them is approved by the individual prior to it being linked with their account. In today’s global organizations where businesses and correspondences are conducted online extensively, it may also be helpful for organizations to provide the necessary IT training to their employees.
The present study addresses a gap in the literature by integrating theories that facilitate novel ways of thinking about the impact of context collapse on loss of face and affect. The multiple yet holistic approach has added to our knowledge about context collapse and the impact this has on loss of face and affect. This study also recommends several practical strategies participants have found useful in helping them prevent breach of privacy or loss of face during context collapse. This is important because we are all increasingly working and communicating in a world that is connected by social media where the issue of context collapse is likely to occur. Thus, having the knowledge and strategy to deal with such collapses is vital for workers.
Footnotes
Appendix A
Appendix B
Dependent measures (items used)
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
