Mastering potential risks:
Employee motivations for self-censorship on social media
Brendon Voogt 10280766 Master’s thesis
Graduate School of Communication
Master Communication Science – Corporate Communication Dr. J.W.M. Verhoeven
ABSTRACT
While social media enable employee voice and stakeholder dialogue, sometimes self-censorship silences employees, because they feel it is too risky to speak out. This cross-sectional survey study among employees (N = 161) aims to unveil the types of risks employees encounter when the wish to express work-related issues on social media such as Facebook, Twitter and Instagram, and what type of self-censorship they employ to reduce these uncertainties. First, based on factor analysis, we found that on external social media two, rather than the four predicted perceived risks can be distinguished: the perceived risk of reputational damage and the perceived risk of publishing low-quality comments. Furthermore, four, instead of seven, self-censorship strategies are deployed by employees: (1) critically reviewing content before publication, (2) tailoring content to imagined audiences, (3) omission of controversial content, and (4) consultation of peers. We found that employees omit controversial content from their messages (i.e., the narrower traditional understanding of self-censorship) to protect personal and corporate reputations. At the same time, they critically review social media content before publication to deal with the risk of harming the quality of conversations. This suggests that altruistic as well as egoistic motives underlie self-censorship in work-related social media use.
Keywords: Social Media, Employees, Risks, Self-Censorship
Mastering potential risks:
Employee motivations for self-censorship on social media
“We can no longer underestimate the impact of social media”. On Augustus 15th, 2014, the Dutch online newspaper NU.nl used this headline to address concern about the growing impact of sharing work-related issues on social media sites with their friends and family as an employee (NU.nl, 2014). The newspaper portrays the image of the ignorant organizational member who does not oversee the potential consequences of such behavior and claims that approximately ten percent of the organizational discharges are attributed to an employee’s negative statements about their job on social networking sites. The article touches upon the increased ability that non-official organizational communicators have to interfere with an organization’s online presences, as is argued by many scholars (Fournier & Avery, 2011; Van Dijck, 2013). It is said that social media enables employees and other stakeholder groups to directly create and share their own view of organizational events (Kaplan & Haenlein, 2010).
An employee’s motive to interfere with such behavior however varies considerably (McDonald & Thompson, 2016). Besides voicing personal observations about their jobs and the organization they work for, work-related disclosure on social media can also serve as a tool to reflect on complex work tasks (Carmeli et al., 2014) or deal with organizational change (Tucker, Yeow, & Viki, 2013). It also offers means to strengthen (online) ties between co-workers and others social actors (Joinson & Paine, 2007), seeing as the feeling of security and approval of peers is considered to be of great concern (Glynn, Hayes, & Shanahan, 1997; Glynn, Hayes, & Shanahan, 2001; Porten-Cheé & Eilders, 2015).
Whereas aforementioned studies stressed beneficial outcomes for this behavior, little is known about what potential risks employees encounter before posting work-related messages on social media and how they cope with these risks by using self-censoring strategies. General studies on organizational silence, however, state that employees give much attention to maintaining a representative self-presentation and try to avoid potential controversy as much as possible (Sleeper et al., 2013). By restraining from voicing their opinion within the organization, employees try to avoid discussions and deal with the pervasiveness of hierarchical structures and authorial figures (Kish-Gephart et al., 2009).
Whether risk perceptions and coping behaviors as self-censorship strategies are present when employees wish to disclose work-related events on their private social media accounts however have yet to be discovered. Although the use of self-censorship strategies on these specific network sites is examined before (Marwick & boyd, 2011), these studies did not relate to an organizational context (Das & Kramer, 2013). Nevertheless, Madsen and Verhoeven (2016) did devote attention to social media usage within an organizational setting, but due to their qualitative research approach, empirical, quantitative evidence still needs to be provided. This study will therefore build upon their research and will examine the relationship between an employee’s perceived risk of disclosing work-related matters on social media and their use of self-censorship strategies through a cross-sectional survey method. Additionally, other factors that might explain the motives for this behavior, such as affective organizational commitment and organizational identification, will be included as well. In this study, the following research question will be answered:
work-related issues on social media and to what extent do they use self-censorship strategies to cope with these potential uncertainties?”
The answers to this research question will not only give use new insight into how, and especially why employees choose to censor themselves on social media, but will also manifest opportunities for organizations themselves. When employees restrain from freely and vocally expressing organizational phenomena, an organization might not be aware of the problems their employees experience. By ensuring an open climate, within and outside of the organization, these insights can be improved (Morrison, 2011).
THEORETICAL FRAMEWORK Work-related social media disclosure
Employees are – along with financers and customers – considered the most influential stakeholder groups for an organization (Fassin, 2009). With their ability to influence the public’s discourse (Verhoeven, 2012), it does not come as a surprise that various researchers call for a deeper scientific understanding of an employee’s motivation to (not) share work-related experiences on social media (El Ouirdi et al., 2015). Verhoeven (2012) defined this online disclosure as a way of organizational expression by employees on non-official communication accounts that are often denoted on one’s own initiative and occurs outside the control of the organization. He furthermore recognizes the importance of an employee’s own perception of organizational phenomena that is shared with in- and out-group members.
Academic research shows that 36.2% of employees engage in work-related disclosure on their private social media accounts (Van Zoonen, Verhoeven & Vliegenthart, 2016) and that this behavior can be explained through social media affordance (Gibbs, Rozaidi, & Eisenberg, 2013). The theory of social media
affordance does not approach social media as a multitude of online functionalities that are cast in stone, but rather focuses on “reciprocal interactions between the social medium, its users and the social context” (Kaufmann & Clément, 2007). According to the theory, social media are thus subject to the meaning an employee gives to each individual medium, which changes across contexts (Treem & Leonardi, 2012). Furthermore, Van Zoonen and colleagues (2014) argued that there are three major motives for employees to engage in work-related social media disclosure. Firstly, they argue that employees use social media to disseminate information with their direct surroundings. In this process, employees share valuable information with in and out group members, but rely heavily on pre-existing corporate information rather than their own personal views of organizational events. Secondly, social media are used as a sensemaking tool. By interacting with organizational actors on social media platforms organizational identities are formed and strengthened (Utz, Schultz, & Glocka, 2013). The third and last motivation is the willingness to express organizational ambassadorship (Van Zoonen, Verhoeven, & Vliegenthart, 2014). An employee who identifies himself with the organization is more willing to engage in work-related matters and it is argued that an employee’s willingness to communicate about its job on social media improves identification processes and self-conforming organizational views (Van Zoonen et al., 2016).
Perceived risk of work-related social media disclosure
Employee social media engagement, however, is not considered to be without risk (Jacobsen & Tufts, 2013; Miles & Mangold, 2014). An unfavorable expression of work-related matters for example can decrease an organization’s reputation among its audience (Mishina et al., 2012). But at the same time negative sentiments can have an
impact on an employee’s reputation (Fournier & Avery, 2011). Kaplan and Haenlein (2010) reported circumstances in which negative disclosure led to decreased career opportunities and, in some cases, even to job loss. A former Microsoft “technical evangelist” for instance was asked to leave the firm after it was reported that he was spreading critical remarks about his employer on several tech blogs (Kaplan & Haenlein, 2010).
It is these career endangerments that have made employees more aware of their role in the social media environment and the potential risk it entails (Child & Petronio, 2011). Hatch and Schultz’s (1997) organizational culture theory provides a possible explanation for this increased risk perception (Sánchez Abril, Levin, & Del Diego, 2012). They argue that organizational members create and share a collective ideology based on self-constructed (unwritten) rules. Employees try to maintain normative professional behavior in order to feel part of the organizational culture. By publishing work-related content on social media, an employee might go against these rules about privacy and disclosure of other sensitive information. Resulting in perceptions that organizational ties will weaken and that this will affect their professional career (Cooren, 2015). Academic literature on impression management (Ashford et al., 1998; Bolino, 1999) furthermore suggest that employees feel the need to present themselves as likeable and competent representatives within and outside the organization (Rosenfeld et al., 1995). Beneficial and professional characteristics are kept in mind before work-related phenomena are shared to avoid negative views of others (boyd & Ellison, 2008).
In terms of internal social media interaction, Madsen and Verhoeven (2016) distinguished four overlapping and interrelated perceived risks that organizational members focus on before they address issues. First, the perceived risk of providing
low-quality posts and comments comprises the idea of sharing irrelevant or inaccurate information with peers. Many social media users are concerned that others will consider their experiences as uninteresting (Stutzman & Hartzog, 2012). Second, the perceived risk of harming one’s own personal reputation is also considered of great concern. This distress concerns anything from antagonistic support to losing professional credibility. Third, the perceived risk to violate set norms and rules is emphasized. Employees especially feel concerned to openly criticize others and lay bare their incompetence. This is in line with Hatch and Schultz’s (1997) aforementioned organizational culture theory, in which they describe that employees try to maintain normative behavior in order to feel part of their organization. The final perceived risk indicates concerns about receiving negative comments and reactions of co-workers and managers. Employees worry that these comments might lead to unwanted critique and will shed light on underlying work-related issues that are intentionally kept under the radar (Morrison, 2011).
Despite these four perceived risks, it must be stressed that potential other dynamics could be at play when employees wish to disclose work-related matters on
external social media, such as Facebook, Twitter and Instagram. However, these
dynamics have not been conceptualized before and empirical quantitative evidence is missing. The present study will therefore build upon Madsen and Verhoeven’s (2016) study on employee risk perception on internal social media, but will provide new insights on the risks employees encounter when they address work-related manners on
external social media.
Self-censorship strategies
use is self-censorship strategy deployment (Das & Kramer, 2013). In a general definition, self-censorship can be described as the act of “withholding relevant ideas
for self-protective reasons” (Detert & Edmondson, 2011, p. 462). A more suitable
definition for the present study can be derived from Madsen and Verhoeven’s (2016) research. The scholars defined self-censorship as the: “act of controlling whether and
how to voice remarks, questions, suggestions or concerns about people, products or processes to others on social media network sites” (p. 389). Academic literature on
organizational and employee silence accredits an important role to the decision calculus employees follow before they decide to speak up or refrain from voicing their personal views on organizational phenomena (Morrison, 2014). Before an employee discloses work-related matters, he will evaluate the perceived efficacy and safety of this behavior. During this process, people thus weigh the effectiveness of their desired results against the potential negative consequences their expression of organizational events might entail (Ashford et al., 1998; Detert & Burris, 2007). This process does not only incorporate real-time, direct human interactions, but also manifests itself in online self-disclosure (Posey et al., 2010). Although this theory is not tested within an organization-related social media context, it is likely that a similar process will occur. Employees who wish to express work-related events on their private social media accounts will in this case evaluate the perceived efficacy – i.e. information dissemination, constructing organizational identities or expressing organizational ambassadorship (Van Zoonen, Verhoeven & Vliegenthart, 2016) – against the potential negative consequences, such as reputation damage and the perceived risk to violate set norms and values (Sleeper et al., 2011). When an employee believes that the perceived risks outweigh the potential benefits, he is more prone to hold silent (Morrison, 2014).
Yet when employees take this proverbial leap of faith and enter the social media area, they are still conscious about potential risks and keep their imagined audiences in mind (Marwick & boyd, 2011). Through privacy setting adjustments on their private social network accounts, employees take an initial step to cope with the uncertainties associated with their imagined audience (Das & Kramer, 2013). Frampton and Child (2013) for instance conclude that 90.5% of the employees approved an online connection with other co-workers, but said they are more vigilant to accept online interaction with organizational actors who are perceived to have authority, due to their ability to harm their professional career. 71.4% of the employees furthermore tighten up their privacy settings and/or block their managers so that they cannot see organizational work-related disclosure (Van Eck Pelechutte et al., 2013). With regards to general “last minute” social media self-censorship, 71% of the Facebook users edited or deleted a published comment at least once (Das & Kramer, 2013) and a majority of the social media users would reduce half of their censored messages if they would be able to control their specific target audiences to not let their messages end up in the wrong hands (Sleeper et al., 2013).
Nonetheless, there is a distinct difference between these self-censorship strategies and those differentiated by Madsen and Verhoeven (2016). Madsen and Verhoeven identified seven major self-censorship strategies that are used by employees on internal social media platforms before an initial message is constructed or published. They propose that employees feel the obligation to contribute in online internal social media messaging, but that each employee has its own way of handling the potential risks they might encounter. Employees who are afraid that their online interactions with colleagues will affect their professional careers, or that they will receive at least negative comments from them, will in most cases decide to
completely refrain from voicing their opinion. Others strategically deploy several self-censorship methods, as expounded in Table 1 (Madsen & Verhoeven, p. 397), to reduce these potential risks. Additionally, employees with a managerial position will deliberate more about their proposed online messages than other organizational actors, because they feel more is at stake. With these self-censorship strategies and their possibility to let an employee cope with his or her uncertainties when disclosing work-related matters on social media, the following hypothesis is formulated: “An
employee’s use of self-censorship strategies in positively influenced by its perceived risk of work-related disclosure on social media” (H1).
Table 1. – Madsen and Verhoeven’s (2016, p. 397) table of self-censoring strategies
Strategy Description Examples
Postpone publishing content. The coworker thinks about it for a while before formulating a post or a comment.
“It goes around in my head, and when I return to my desk half an hour later, then I have found the right phrase I want to use, so that I comment in the right way.” Phrase or reframe content
differently.
Writing, reading, and rewriting carefully before posting.
“I write directly [on ISM], but it takes a long time to write it, if it is 10–15 lines.”
Imagine responses from
organizational members. Trying to imagine responses from the audience and writing a post or a comment to meet or avoid the responses.
“This color has been changed from yellow to blue. And then they ask why? I might just as well give them the answer straight away, because there will always be a why, or I write: There is no why?”
Ask for a second opinion. The coworker asks for a second opinion from a manager or colleagues.
“I am more inclined to discuss it with my colleagues around the table.”
Choose another channel. Talking with colleagues instead of airing a frustration or an opinion on ISM.
“The written word has its limitations, and sometimes it is better to engage in a dialog.” Withdraw. Deciding not to post or comment. “Don’t want to say too much. I
have also heard colleagues say that. You do have a desire to comment, but you don’t do it.” Write only positive comments. Only posting harmless, positive
comments.
“I might write something like ‘Well done.”
Affective organizational commitment
Finding organizational constructed normative rules important has a regulatory effect on the willingness to express work-related manners (Morrison & Milliken, 2000). This organizational commitment is defined by Weiner (1982, p. 481) as “the totality
of normative pressures to act in a way which means organizational goals and interest”. Meyer and Allen (1991) furthermore emphasize affective devotion to the
organization, the perceived benefit to work at the organization and the potential costs of leaving as contributing factors.
Organizational commitment studies show that employees value an open organizational climate in which questions and remarks can freely be discussed (Vokalo & Bouradas, 2005; Panahi, et al., 2012). However, high-committed employees are concerned that vocal expressions of organizational events might harm their professional reputation. In order to reduce these risks, committed employee will more easily refrain from expressing work-related issues within the organizational environment than less committed employees (Dedahanov & Rhee, 2015). Since the present study focuses on the expression of work-related issues on social media, it is thus expected that highly organizationally committed employees are more likely to use self-censorship strategies on social media when they perceive that this might affect the organization they work for than those who are less committed. The second hypothesis reads as follows: “Affective organizational commitment will moderate the
relationship between the perceived risk to disclose work-related matters on social media and the use of self-censorship strategies to cope with these uncertainties (H2).
Organizational identification
subject to constructed organizational normative rules, but rather focuses on “self-definitional aspects of organizational membership” (Van Knippenberg, 2006). Mael and Ashford (1992, p. 109) defined this concept as “the perception of oneness with or
belongingness to an organization, where the individual defines him or herself in terms of the organization(s) of which he or she is a member”. This definition is derived
from the social identity theory, which states that individuals classify themselves and others into categories based on specific characteristics (e.g. gender, age, religion, family or the organization they work for) to get a better understanding of the world (Turner et al., 1987).
High-identifying employees take pride in the successes of their organization and share these positive attitudes on regularly basis (Van Knippenberg & Van Schie, 2000; Riketta, 2005). This sense of pride and support contributes to an employee’s awareness of organizational corporate reputation (Helm, 2011). However, in accordance to academic literature, when corporate reputation is in dispute, high-identifying employees invest personal effort to reduce these negative sentiments through increased performance effectiveness and task involvement (Efraty & Wolfe, 1988). In reference to the present study, it is thus expected that highly identifying employees will do their utmost best to protect the organization’s reputation trough self-censorship deployment on social media. Therefore, the third hypothesis reads as follows: “Organizational identification will moderate the relationship between the
perceived risk of disclosing work-related content and the use of self-censoring strategies” (H3).
METHOD
This chapter will elaborate on the study design, study sampling, data collection and measurement instruments that are used for this research. In order to examine the
relationship between an employee’s perceived risk of disclosing work-related content and its use of self-censorship strategies to cope with these uncertainties, a quantitative cross-sectional survey was distributed. Due to its ability to measure a great amount of data about many characteristics, a survey design method is a preferred research technique to study correlations between several variables (Flick, 2014). Additional to main analyses examinations, potential moderation effects of organizational commitment and identification are examined.
Sampling and data collection
The questionnaire was distributed to the target respondents from 15 November 2016 until 27 November 2016. To meet the study’s set goals, a number of criteria were compiled. First, for ethical reasons, respondents must be older than 18-years to participate. Furthermore, since this study focuses on employees as the target group, respondents must work in an organization at least 12 hours a week, the minimal hour criteria that is used by the Dutch Central Agency for Statistics (2016) to be considered as part of the population’s workforce. Besides these demographical characteristics, respondents must have a social media account (e.g. a Facebook, Twitter or Instagram account) and must login at least once a month on one of these accounts to be included for this research. There were no restrictions regarding organization or job type.
Questionnaire distribution took place through convenience sampling. First, organizational HR and/or communication departments were approached. The main goal of this study was explained through e-mail and permission was asked to distribute the survey among their employees. Because the anticipated response rate was lower than expected, the survey was subsequently distributed through various social media channels. When an employee volunteered to participate, an introduction
message was send out to them. The message explained that the questionnaire was part of a Master’s thesis for the program Communication Science – Corporate Communication of the University of Amsterdam. During and after the survey, absolute anonymity was ensured. In addition to this, it was made clear that respondents could contact the master’s student and his thesis supervisor any moment in time with questions and remarks about the study. An informed consent form that had to be approved by the respondents ensured the ethical guidelines of the Amsterdam School of Communication Research (ASCoR, 2013).
In total, 312 participants volunteered to participate. From this group, 230 respondents finalized the whole questionnaire. 60.4% of the participants identified themselves as female, 39.1% as male and one person did not specify a gender, with an average age of 34.2 years (SD = 11.79). The educational level of the participants was divided as follows: 43.9% of the volunteers have obtained their academic (under)graduate degree, 25.7% finalized a Dutch hbo program, 10% succeeded to obtain a Dutch vwo-diploma and the rest of the respondents were divided over different educational programs. The respondents furthermore indicated that they work 33-hours a week on average. 17% of the sample works in education, 13.9% in the media and journalism sector, 10% as communication professionals and the rest in other domains. In terms of social media usage, Facebook can be considered as the most frequently used social media site to disclose personal experiences with a usage rate of 85.7%. Instagram, Twitter and Snapchat follow shortly with respectively 53.9%, 43.9% and 32.6%. Most participants publish messages on social media 1 to 4 times a month (23.9%). Work-related disclosure on social media occurred in 70% of the cases (n = 161), with Facebook (48.3%) and LinkedIn (42.2%) as the most frequently used sites. The respondents however published less work-related content
than personal matters: 1 to 10 times a year on average. Since we are only interested in employees who disclose work-related events on their private social media accounts, only this specific group was used for data analyzes.
Measurement instruments
During the questionnaire, respondents were asked to indicate their perceived risk to share work-related matters on social media and which self-censorship they use to cope with these potential risks. Thereafter, this research examined to what extend the relationship between the perceived risk of this disclosure and the use of self-censorship strategies is moderated by affective organizational commitment and organizational identification. These individual variables are operationalized as follows:
Perceived risk of work-related disclosure. The measurements for the
perceived risk of work-related disclosure is developed and derived from Madsen and Verhoeven’s (2016) research on internal social media self-censorship strategies of employees. In their qualitative study, structural interviews were held with bank employees and they were asked about the potential risks they encounter when they disclose work-related matters on internal social media. In their conclusion, the researchers acknowledged four different perceived risks mentioned by the bank employees. From these four risks and the corresponding employee statements, a self-constructed Likert scale was developed. This scale consisted of 34 items which could be answered from 1 = “strongly disagree” to 7 = “strongly agree”. Example items were: “When I want to publish something about my work on social media, I am afraid that I will address irrelevant topics”, “When I want to publish something about my work on social media, I am afraid to lose credibility” and “When I want to publish something about my work on social media, I am afraid about my co-workers’ reaction” (Please see Appendix D for a full list of items). Unlike Madsen and Verhoeven’s
(2016) study, principal component factor analyses (PCA) with Varimax rotation indicated five independent components with an Eigenvalue above one. However, scree plot analyses and variance percentage distribution illustrated two main components, respectively the perceived risk of reputational damage and the perceived risk of publishing low-quality comments, that could be constructed as independent variables for this research. Each item that is used to construct the individual variables and their corresponding factor loadings (above .45) can be found in Table 2, along with the variable’s explained variance and its Cronbach’s Alpha value.
Self-censorship strategies. Just like the independent variable, Madsen and
Verhoeven’s study (2016) form the measurement basis for the dependent variable(s). A self-made Likert scale with a total of 30 items was developed from this study. The items could be answered from 1 = “strongly disagree” to 7 = “strongly agree”. Example items were: “Before I publish something on social media about my work on social media sites I sometimes take time to deliberate”, “Before I publish something on social media about my work on social media sites I ask my colleagues to review my post” and “I choose to formulate an intended comment positively to not hurt anyone” (see Appendix D). Principal component factor analyses (PCA) with Varimax rotation indicated a total of seven components with an Eigenvalue above one, which corresponds with Madsen and Verhoeven’s study. However, scree plot analyses and variance percentages distribution indicated a total of four variables that could be constructed to form the dependent variables that measure self-censorship strategies. Each item that is used to construct the individual variable and its corresponding factor loadings (above .45) can be found in Table 2, along with the variable’s explained variance and its Cronbach’s Alpha value.
Table 2. Scale measurements for independent, dependent and moderator variables
Construct and indicators Loadings Eigenvalue EV* α
Perceived risk of reputation damage (PRD)
“When I want to publish something on social media about my work I am afraid that…”
16.176 23.4% .952
…I will complain too much. .75
…I will bring my job in discredit. .75 …My colleagues will not appreciate my comment. .75 …I will damage the professional image others have about me. .73 …It will have negative consequences for my career. .72
…I will violate unwritten rules. .69
…My manager will not appreciate my comment. .68 …I will receive comments from others. .68 …My comment will have effect on my position within the organization. .65 …Others will judge about my competences. .63 …I will reveal too much about myself. .60 …My comment will result in a long discussion. .59
…It will not worth it. .52
…I will write something that will not be appreciated by others. .48 Perceived risk of publishing low-quality messages (PLQ)
“When I want to publish something on social media about my work I am afraid that…”
3.290 13.4% .900
…My comment will not be valuable for others. .80 …My comment will not be interesting enough. .79
…I will waist people’s time. .78
…My comment will not be unique enough. .60 …My comment will be of poor quality. .58 …I will bring up irrelevant topics. .47 …My comment will not be concrete enough. .45 Critically reviewing comments before publishing (CRC)
“Before I publish something about my work on social media…”
7.239 16.8% .890
…I sometimes take time to carefully formulate my comment. .87 …I sometimes make sure that I use the right words to let my message come
across as I intended. .81
…I sometimes try to prevent misunderstandings. .79 …I sometimes make sure that my message comes across as professional. .78 …I sometimes take time to deliberate about the potential consequences of my
message .75
…I sometimes postpose the process to critically review my comment. .66 …I sometimes make sure that I will not be carried away with my feelings. .63
NOTE: *EV = Explained variance.
Construct and indicators Loadings Eigenvalue EV* α Consulting peers (CP)
“Before I publish something about my work on social media…”
4.461 15.6% .876
…I sometimes walk to my colleagues first to discuss the subject of my
message. .77
…I sometimes ask my colleagues to provide feedback on my intended
message. .72
…I sometimes discuss the subject of my message first during lunch. .72 …I sometimes ask my manager to provide feedback on my intended message. .71 …I sometimes try to talk to others about the subject of my message first. .70 …I sometimes ask my friends to provide feedback on my intended message. .69 …I sometimes discuss the subject of my message with my immediate
surroundings first via e-mail. .69 …I sometimes discuss the subject of my message with my immediate
surroundings first via telephone. .67 …I sometimes ask my family to provide feedback on my intended message. .62 Omissions of controversial content (OCC)
“Before I publish something about my work on social media…”
2.655 9.5% .805
…I sometimes decide not to publish an intended comment to stay out of
trouble. .81
…I sometimes decide not to publish an intended comment to avoid
controversy. .80
…I decide to positively formulate my message to not hurt others .64 …I decide to positively formulate my message to avoid criticism. .59 Tailoring content to imagined audience (TCA)
“Before I publish something about my work on social media…”
1.892 8.6% .722
…I sometimes make sure that my message suits the expectation from my colleagues.
.75 …I sometimes try to image what questions others have about this. .69 …I sometimes try to formulate my message so that it will result in a
discussion.
.64 …I sometimes try to explain my motives as good as possible so that everyone will understand me.
.51
Affective organizational commitment (AOC) 3.235 53.9% .808
I do not feel like “part of the family” at my organization (reversed). .87 This organization has a great deal of personal meaning for me. .84 I do not feel emotionally attached to my organization (reversed). .83 I do not feel a strong sense of “belonging” to my organization (reversed). .67 I would be very happy to spend the rest of my career with this organization. .60 I really feel as if this organization’s problems are my own. .52
Organizational identification (OI) 3.284 54.7% .827
When someone praises the organization I work at, it feels like a personal
compliment. .86
The organization’s successes are my successes. .83 When I talk about my work, I usually say ‘we’ rather than ‘they’. .71 I am very interested in what others think about the organization I work at. .67 When someone criticizes the organization I work at it feels like a personal
Affective organizational commitment (moderator). The moderating
variable affective organizational commitment was measured by means of a pre-existing six-item scale of Meyer, Allen and Smith (1993). The respondents were asked to answer the six items on affective organizational commitment on a seven-point Likert scale, ranging from 1 = ‘strongly disagree’ to 7 = ‘strongly agree’. All used items and their corresponding factor loadings (above .45), as well as the variable’s explained variance and its Cronbach’s Alpha value, can be found in Table 2.
Organizational identification (moderator). To measure an employee’s
organizational identification, Mael and Ashforth’s (1992) six-item was used. The respondents were asked to answer the six items that collectively formed the organizational identification variable on a seven-point scale, ranging from 1 = ‘strongly disagree’ to 7 = ‘strongly agree’. All used items and their corresponding factor loadings (above .45), as well as the variable’s explained variance and its Cronbach’s Alpha value, can be found in Table 2.
RESULTS
In general, descriptive analyses (Table 3) indicate that an employee’s perceived risk of reputational damage and the risk of publishing low-quality comments is just below mid-point on a seven-point scale before they address work-related issues on social media. However, from the four self-censoring strategies that are distinguished through factor analyses, we can state that employees constantly critically review their constructed message before sending them into the online arena. The respondents furthermore reported that they every so often withhold themselves from sending controversial messages and tailor their content to their imagined audience. To call in peers before a comment is posted on the other hand is a less widely used strategies.
Table 3 – Descriptives independent and dependent variables
Main analyses
In order to test whether any statically significant relationship between an employee’s perceived risk and its use of self-censorship strategies is present, multiple linear regression analyses are performed for each individual dependent variable. Besides measuring associations between these independent and dependent variables, the demographic control variables age, gender, educational background, job position and average work week hours are included as well. Since gender and job position are categorical variables, dummy variables are created in order to interpret them as continuous. The ordinal variable educational background is furthermore split on its median (Mdn = 6.00, Hbo) in high and low education. From this median split another dummy variable is created. All variables are tested on the assumptions of regression (e.g. linearity and multicollinearity) before calculating the models (Field, 2013). These assumptions are met.
Critically reviewing content. The multiple regression model for the
self-censorship strategy critically reviewing content is considered to be significant and is thus able to predict the relationships with this specific dependent variable. However, with 10.0% of the variation within critically reviewing content, explained by the
M SD N
Perceived risk of reputation damage (PRD) 3.04 1.30 161 Perceived risk of publishing low-quality comments (PLQ) 3.00 1.23 161
Critically reviewing comments before publishing (CRC) 5.45 1.04 161
Consulting peers (CP) 2.57 1.09 161
Omission of controversial comments (OCC) 4.09 1.38 161 Tailoring content to imagined audience (TCA) 4.14 1.22 161
independent and control variables, the predication is considered rather small (see Table 4). The main effects expounded in Table 4 furthermore indicate a marginal significant association between an employee’s perceived risk of publishing low-quality comments and it’s use to critically review the constructed message, b* = .19,
t = 1.97, p = .051, CI [.00, .32]. This moderately strong effect shows that for each
additional point on the scale for the perceived risk of publishing low-quality comments, the average propensity to critically review a constructed work-related message increases with .16. The control variable job position (manager versus employee) furthermore indicates a moderately strong significant relationship with critically reviewing content, b* = -.28, t = 3.53, p = .001, CI [-.97, -.28] (Table 4). A manager’s average propensity to critically review its message is .38 more than non-managers.
Consulting peers. The multiple regression model for consulting peers is also
considered to be significant, as can be seen in Table 4. Yet, only 9% of the proportion of the variation within consulting peers is explained by the independent and control variables, which makes the predication small as well. Nonetheless, a marginal significant relationship is found between the perceived risk of reputation damage and consulting peers, b* = .19, t = 1.97, p = .051, CI [.00, .32]. This moderately strong effect indicates that for each additional point on the scale for the perceived risk of reputation damage, the average propensity to consult peers increases with .16 (Table 4).
Omission of controversial content. The third regression is also found to be
significant for the self-censorship strategy omission of controversial content (Table 4). Unlike previously mentioned regression models, this model has a moderately strong predication (R2 = .30). Yet, only one significant relationship is found. The multiple
regression indicates a strong association between the perceived risk of reputation damage and the omission of controversial content by employees, b* = .53, t = 6.36,
p > .001, CI [.37, .71]. The regression indicates that for every additional point on the
scale for perceived reputation damage, the average propensity to use the self-censoring strategy omission of controversial content increases with .51 (Table 4). Tailoring content to imagined audience. The final self-censorship strategy
tailoring content to an employee’s imagined audience did not form a significant
multiple regression model with the two independent variables and the additionally added control variables. The model is thus not able to predict the associations between these variables and the individual regressions could therefore not be interpreted (Table 4).
Hypothesis confirmation. The first hypothesis read as follows: “an
employee’s use of self-censorship strategies in positively influenced by its perceived risk of work-related disclosure on social media (H1)”. Above multiple regression
analyses indicate several marginal and significant relationships between the perceived risk scales and the four different censorship strategies. However, not all four self-censorship strategies formed a significant association with these variables. Hypothesis 1 can thus only be partially confirmed.
Moderation analyses
Affective organizational commitment and organization identification were included in order to test the second and third hypothesis. It is expected that both concepts will positively moderate the relationship between the perceived risk of online work-related disclosure and an employee’s use of self-censorship strategies. In order to test these
Table 4.
Regression models to predict dependent variables.
Note: PRD = Perceived risk of reputation damage, PLQ = Perceived risk of publishing low-quality comments. Critically reviewing content Consulting peers Omission of controversial content Tailoring content to imagined audience b* p F R2 b* p F R2 b* p F R2 b* p F R2 - .002 3.48 .10 - .003 3.27 .09 - < .001 10.84 .30 - .068 1.93 .04 Constant - < .001 - <.001 - < .001 PRD -.04 .715 .19 .051 .53 < .001 PLQ .19 .051 .17 .073 .06 .478 Age .14 .073 -.02 .853 .05 .450 Gender .06 .464 .07 .389 .06 - .384 Education -.01 .867 -.17 .101 .06 - .347 Job position -.28 .001 -.11 .157 .04 .524 Average work week hours -.03 .668 -.02 .774 .00 .976
effects, an interaction term between each moderator (affective organizational commitment and organizational identification) and the two independent variables is created. Almost all multiple regression models were found to form a significant model and were thus able to predict the relationships of the four dependent variables. Only one regression analysis, in which the interaction term between the perceived risk of reputation damage and organizational commitment (PRD*OC) was included, did not form a significant model (Table 5 & 6, Appendix A)
Affective organizational commitment. Table 5 shows a 3% increase of the
proportion of the variation within critically reviewing content after the moderator affective organizational commitment and the interaction term PRD*OC were included. Main analyses initially indicated a marginal relationship between the perceived risk of reputation damage and using the self-censorship strategy critically reviewing content (see Table 3). However, after including the moderator and interaction variable, the value for significance drops under the threshold (p = .05), and becomes significant,
b* = .61, t = 2.19, p = .034, CI [.04, .90]. Besides this change, a strong negative
significant relationship is found between the interaction term and the dependent variable critically reviewing content, b* = -.73, t = -2.36, p = .020, CI [-.20, -.02]. The average propensity to critically review content furthermore decreases .72 on the interaction scale, meaning that employees who are highly committed to the organization, will less likely critically review their content when the perceive that this could risk their reputation (see Figure 1). Because hypothesis 2 states that highly committed employees will critically review their content more when they perceive reputational damage than low-committed employees, and no other moderation significant relations are found, hypothesis 2 will be rejected.
Figure 1 – Affective organizational commitment as moderator for the relationship perceived risk of reputation damage and critically reviewing content.
Note: PRD: perceived risk of reputation damage, AOC: affective organizational commitment. Organizational identification. Two moderating effects are found for the
relationship between both independent variables on the dependent variable tailoring content to the imagined audience. With an increase of respectively two and three percent of the variation proportion within the variable tailoring content to the imagined audience, after the moderator organizational identification and the interaction term PRD*IO and PLQ*OI were included, the prediction must still be interpreted as small. However, the interaction terms PRD*OI, b* = -.92, t = -2.77, p = .006, CI [-.29 -.05], and PLQ*OI, b* = -.81, t = -2.17, p = .031, CI [-.27, -.01] form a strong association with the dependent variable (see Table 7 & 8, Appendix A). The average propensity to tailor content to the imagined audience decreases with .17 when an employee perceive that this might result in reputation damage (see Figure 2) and with .14 when he or she is concerned to construct low-quality messages (see Figure 3), indicating that employees who do not identify themselves with the organization, will more likely tailor their content to their imagined audience when they perceive that this will result in the risk of reputational damage and publishing low-quality comments. We can thus state that hypothesis 3 will be rejected.
1 2 3 4 5 6 7 Low PRD High PRD Cr it ic al ly Re vi ewi ng Co nt en t Low AOC High AOC
Figure 2 – Organizational identification as moderator for the relationship perceived risk of reputation damage and tailoring content to imagined audience.
Note: PRD: perceived risk of reputation damage, IC: organizational identification.
Figure 3 – Organizational identification as moderator for the relationship perceived risk of publishing low-quality comments and tailoring content to imagined audience.
Note: PRD: perceived risk of reputation damage, IC: organizational identification. CONCLUSION
This study is one of the first in the field of communication and other social sciences to examine the types of risks employees perceive before they disclose work-related messages on social media and makes a distinction between several different self-censorship strategies that could be used to cope with these uncertainties. By providing quantitative results, the present study contributes to new empirical understanding of
1 2 3 4 5 6 7 Low PRD High PRD T ai lor in g con te n t to im agi n ed au d ie n ce Low OI High OI 1 2 3 4 5 6 7 Low PRD High PRD Ta il or in g co nt en t to im ag in ed a ud ie nc e Low OI High OI
these characteristics. A new developed framework highlights the differences between external and internal social media use, the latter as expounded by Madsen and Verhoeven (2016), and can be used for others that would like to examine employee social media behavior of employees. Through the following research question development, an attempt to give a substantive answer to aforementioned points of interest was made: “What kind of perceived risks do employees encounter when they
want to address work-related issues on social media and to what extent do they use self-censorship strategies to cope with these potential uncertainties?”.
Before this question could be answered, measurement instruments had to be developed in order to test an employee’s risk perception and self-censorship strategy use. A small number of differences are found between the risks distinguished by Madsen and Verhoeven (2016) and the present study. This study’s factor analyses indicated two key perceived risks, respectively the perceived risk of reputational damage and the perceived risk of publishing low-quality comments, but did not find the clear distinction in concerns about violating norms and rules and receiving (negative) comments from co-workers that Madsen and Verhoeven (2016) did. A possible explanation is attributable to the intrinsic differences between internal and external social media. Internal social media enables contributors to take cues on normative behavior much faster than external social media sites do, since boundaries between professional and private identities on internal social media are much clearer (Ollier-Malaterre et al., 2013; Uysal, 2016;). Furthermore, with regards to self-censorship, four self-censorship strategies are found: (1) critically reviewing content, (2) consulting peers, (3) omission of controversial comments, and (4) tailoring contents to an imagined audience. Madsen and Verhoeven (2016) initially proposed seven individual self-censorship strategies. The present study’s framework suggests
that these strategies are still present, but some of them together form coherent latent variables.
Subsequently, five contributing conclusions can be drawn from this research. The most prominent finding addresses the way employees deal with potential reputational risk. When an organizational member perceives that he crosses an important self-imposed boarder by damaging his own and the organization’s reputation while writing a work-related message, aimed to be published on social media, the employee will in most cases either decide to refrain completely from voicing his opinion or will resolve this matter through positive message framing. This omission of controversial content appears to ease the concerns of hidden threats, like diminishing career opportunities or receiving unwanted critique from peers (Fournier & Avery, 2011). As conceptualized by Van Dyne and colleagues (2003), this so-called defensive and prosocial silence withholds employees from expressing controversial comments in order to protect their reputation and the comments’ potential negative consequences (Pinder & Harlos, 2001). The basic assumption for this theory suggest that employees seek means to avoid personal discomfort by actively omitting negative statements (Rosen & Tesser, 1979).
Nevertheless, controversial message avoidance is not the only used self-censorship strategy to reduce potential reputational damage. Even though this strategy is used to a lesser extent than controversy avoidance, an employee’s use of content tailoring is subject to the way he identifies himself with the organization. It can be concluded that low-identifying employees will only tailor social media messages in order to fit the expectations of their imagined audience when they perceive that their own reputation is at risk. This egocentric behavior ties in with social media’s key utilizations: the construction of a valuable self-presentation (Sleeper et al., 2013).
Although low-identifying employees initially are not led by organizational interests (Dutton, Dukerich, & Harquail, 1994), when these employees perceive that their own self-presentation could be compromised, due to work-related social media disclosure, they take better account of their target audiences and construct a favorable self-image (Marwick & boyd, 2011). Meanwhile, high-identifying employees do not perceive potential reputational damages as a motive for content tailoring, because they already create more favorable work-related messages about the organization in order to express their sense of belonging (Van Knippenberg & Van Schie, 2000).
Furthermore, there is a moderating relationship between an employee’s organizational commitment and its deployment of a critical reviewing technique to reduce potential reputational damages. Contrary to expectations, reputational risk perceptions do not motivate highly committed employees to critically review their own content. The present study’s findings show an opposite trend: employees who indicated that they are very committed to their organization, are less likely to review their initially constructed message on potential misunderstandings that could put their reputation at risk. In agreement with academic research, highly committed employees will utilize controversy avoidance techniques in any case to reduce these risks (Tangirala & Ramanujam, 2008). When these organizational members perceive that their own and the organization’s reputation is at risk, they will in this case thus move onto controversy avoidance rather than critically reviewing their content.
Beside the perceived risk of reputational damage, a second potential risk is distinguished: the concern to produce low-quality messages. Unlike the first risk, only two self-censorship strategies - critical content reviewing and content tailoring - are used to cope with this uncertainty. Although the relationship between content quality and the use of critically reviewing techniques is marginally significant, there seems to
be a positive trend between these two variables. Since a significant association is formed after including the interaction of reputational damage and critical reviewing strategies, we feel confident enough to state that organizational members, and especially those with a managerial position, will examine their posts on potential shortcomings to reduce their concern of poor quality disclosure. In an online environment where the imagined audience is not always clear (Marwick & boyd, 2011), critically reviewing strategies enables employees to control their emotions and come across as being controlled (Grant, 2013), which can lead to greater social capital (Ellison et al., 2007). It can furthermore be concluded that an employee’s use of content tailoring to reduce potential low-quality comments is subject to their organizational commitment. Just as the reputational damage risk, we can state that non-organizational identifying employees will only tailor their messages to fit their imagined audience (Marwick & boyd, 2011) views when they perceive a high possibility of their message being considered as of poor quality. At the same time, high-identifying employees do not perceive content tailoring as a useful way to reduce poor quality comments, since they already take account for potential risks when disclosing organizational matters, as mentioned before (Van Knippenberg & Van Schie, 2000).
Finally, the present study found states that peer consultation is a less widely used strategy to reduce risk. Asking for feedback is not always an easy thing to do (Lucas et al., 2016). When people are unsure about how others might react to their personal concerns, such as experienced risks, they are less likely inclined to discuss these matters with people in their direct surroundings (Wheeless & Grotz, 1977).
DISCUSSION!
The present study addresses several interesting points. First, the concerns about reputational damages and the construction of low quality comments ties in with the impression management theory. This theory states that employees wish to come across as competent and likeable organizational actors (Yun, Takeuchi & Liu, 2007). Before addressing work-related issues, employees take concerns about normative behavior violation into account in order to maintain the favorable image others have of them (Hatch & Schultz, 1997). Furthermore, the absence of a clear distinction between an employee’s professional and organizational reputation risk, as distinguished by Madsen and Verhoeven (2016), can possibly be explained by an employee’s imagined audience (Marwick & boyd, 2011). Contrary to internal communication platforms, it is not always clear to whom messages on (external) social media sites are addressed (Stutzman et al., 2012). Anyone with or without a direct relation to the social media user can possibly access an employee’s work-related social media message. Therefore, employees try to construct content that is appropriate for everyone. Concerns about low-quality comments can thereby be described based on social media’s basic principles: social capital construction (Ellison et al., 2007). Social media allow its participators to construct and strengthen relationships that are unavailable in the offline world (Williams, 2006). However, before these processes can occur, clear and unambiguous content is needed in order to create better social capital and to form strong relationships (Valenzuela, Park & Kee, 2009). In this specific case, employees are concerned that potential poor-quality comments and its indirect intentions will be mistaken by others and this will affect the creation of social capital.
Subsequently, the present study’s expounded self-censorship strategies can be linked to the expectancy theory (Lawler & Suttle, 1973). This theory explains an employee’s motivation to choose a certain behavioral option over another and consists of three key components: behavioral expectancy, instrumentality and valence. Focusing on the present study, employees who practice self-censorship strategies on social media in order to reduce their risk concerns (expectancy), believe that their behavior will result in the preservation of reputations and construction of well-evaluated content (reward). However, when an employee does not consider these risks as important (value), he will not easily deploy self-censorship strategies and take the perceived risks for granted (Lawler & Suttle, 1973). Nevertheless, although many scholars argue that social media use is rather thoughtless and impulsive (Van Zoonen, Verhoeven & Elving, 2014), this study shows that employee are well aware of the risks work-related disclosure entails and how they can manage these risks through self-censorship strategy deployment. This behavior shows that there are some existing advance social media skills (Novakovich, Miah, & Shaw, 2017) with which employees show that they are capable to adjust their behavior in complex situations and are able to take social etiquette into consideration (Livingstone, 2014).
The final important point of discussion specifically focuses on the variables of affective organizational commitment and organizational identification. Contrary to what was expected, these organizational characteristics did not positively moderate the relationship between an employee’s perceived risks and its behavior to critically review content. In fact, high-identifying and committed employees did not show any difference in self-censorship use or even decreased their censorship strategies to overcome potential risks. An explanation for this could be that this specific group
always takes the potential consequences for themselves, and especially the organization, into account (Morrison, 2014).
Managerial implications
The question arises whether the implementation of self-censorship strategies on social media is beneficial for organizations and their personnel. The demonstrated presence of risk perception and responsible social media use show a fair amount of digital literacy, in which personal and professional self-harm is reduced to its minimum (Lankshear & Knobel, 2008). Despite this assumption, this study shows that many employees in fact are concerned about reputational damage when they address work-related issues on social media and perceive that they will be judged when they would publish comments that are not up to the norm. By means of self-censorship strategy, employees try to get a grip on these issues. However, the present study also demonstrated that there is a group of employees who do not identify themselves with the organization and perceive little risks during work-related disclosure on social media, making themselves and the organization’s brand image vulnerable to others. Through constructive social media training managers, and others in executive positions, can make these employees more aware of their behavior and reduce any negative impact. Besides, the creation of an open organization climate in which an employee can voice personal opinions – without any restrictions – on social media can enhance organizational ties and beneficial corporate views among employees (Kernstock & Brexendorf, 2009), in which opportunities arise to turn them into online brand ambassadors (Verhoeven, 2012).
Limitations
This research is not exempt from some limitations. Primarily, a convenience sampling technique was used to carry out this study. This technique enables us to reach a great number of possible contributors, but, on the other hand, limits us from generalizing the outcomes. Thereby, the sampling itself was not distributed equally in terms of field of work. These was an overpopulation of employees working in the communication, media and journalistic sector. Since social media generally plays an important role in the execution of their job tasks (Lariscy et al., 2009), people working within these labor fields might be more aware of the possible consequences social media use entails than others. Moreover, as pointed out by Flick (2013), cross-sectional quantitative survey designs are a useful way to measure many characters at once, but at the same time causality and long-term effects are ruled out. Furthermore, due to this study’s explorative character, measurement scales had to be developed to understand the different types of risks employees encounter when they wish to disclose work-related issues on social media and what kind of self-censorship strategies they use to reduce these uncertainties. Although these scales form an important conceptual contribution to scientific research - as pointed out in the results section - future research will need to validate these measurements. Structural equation modeling for instance would be useful due to its confirmatory technique, to detect any possible flaws and which would seek unbiased estimates within and between the several types of distinguished variables. Concerning the present study’s purpose, other (media) channels that possibly contribute to work-relate disclosure were ruled out, giving us an unclear view of an employee’s use and perceived risks between these channels. Future studies should examine the possible expected gains from self-censorship strategies deployment. Finally, it was beyond the scope of this study to
examine to what extent different organizational tendencies, such as organizational change and ambassador citizenship behavior, have effect on an employee’s perceived risk of online work-related disclosure and his use of self-censorship strategies.
REFERENCES
ASCoR/Amsterdam School for Communication Research (2013). Ethical review procedure for the department of communication science, University of Amsterdam.http://student.uva.nl/binaries/content/assets/studentensites/
fmg/communication-science-masters/ethical-review/manual_full_procedure for-ethical-review_cw-english-version.pdf?2829351267706.
Ashford, S. J., Rothbard, N. P., Piderit, S. K., & Dutton, J. E. (1998). Out on a limb: The role of context and impression management in selling gender-equity issues. Administrative Science Quarterly, 23-57.
Bolino, M. C. (1999). Citizenship and impression management: Good soldiers or good actors?. Academy of Management Review, 24(1), 82-98.
boyd, d. m., & Ellison, N. B. (2008). Social networking sites: definition, history and scholarship. Journal of computer-mediated communication, 13, 210-230. Carmeli, A., Sheaffer, Z., Binyamin, G., Reiter‐Palmon, R., & Shimoni, T. (2014). Transformational leadership and creative problem‐solving: The mediating role of psychological safety and reflexivity. The Journal of Creative
Behavior, 48(2), 115-135.
Child, J. T., & Petronio, S. (2011). Unpacking the paradoxes of privacy in CMC relationships: The challenges of blogging and relational communication on the internet. Computer-Mediated Communication in Personal Relationships, 21-40.
Cooren, F. (2015). Organizational discourse: Communication and constitution. John Wiley & Sons.
Das. S., & Kramer, A. D. (2013). Self-Censorship on Facebook. In ICWSM.
Dedahanov, A. T., & Rhee, J. (2015). Examining the relationships among trust, silence and organizational commitment. Management Decision, 53(8), 1843 – 1857
Detert, J. R., & Edmondson, A. C. (2011). Implicit voice theories: Taken-for-granted rules of self-censorship at work. Academy of Management Journal, 54(3), 461–488. doi:10.5465/AMJ.2011.61967925
Detert, J. R., & Burris, E. R. (2007). Leadership behavior and employee voice: Is the door really open?. Academy of Management Journal, 50(4), 869-884.
Dutch Central Agency for Statistics (2015). Beroepsbevolking, 12-uursgrens. https://www.cbs.nl/nl-nl/onzediensten/methoden/begrippen?tab=b#d= beroepsbevolking--12-uursgrens—
Dutton, J. E., Dukerich, J. M., & Harquail, C. V. (1994). Organizational images and member identification. Administrative science quarterly, 239-263.
Efraty, D., & Wolfe, D. M. (1988). The effect of organizational identification on employee affective and performance responses. Journal of Business and
Psychology, 3(1), 105-112.
Ellison, N. B., Steinfield, C., & Lampe, C. (2007). The benefits of Facebook “friends:” Social capital and college students’ use of online social network sites. Journal of Computer!Mediated Communication, 12(4), 1143-1168.
El Ouirdi, A., El Ouirdi, M., Segers, J., & Hendrickx, E. (2015). Employees' use of social media technologies: A methodological and thematic review. Behaviour
