University of Amsterdam
Graduate School of Communication: Corporate Communication Science
Master’s Thesis
Anonymous and Pseudonymous Brand-Related
Communications on Social Media: The Effects on Toxicity,
Transparency and Self-Disclosure
Zygimantas Valiulis (12268895)
Supervisor: Dr. S.C. (Suzanne) de Bakker Date of Submission: 28/06/2019
Abstract
Online Disinhibition theory suggest that computer-mediated communications (CMC) allow users to be more toxic and self-disclosing. However, a huge number of overall online communications are made by anonymous and pseudonymous users, which can have an even bigger effect on users’ online disinhibition. The current research applied grounded theory approach and qualitative content analysis to conduct an in-depth examination of how varying levels of user' identification (anonymous, pseudonymous & identified) might be influencing their CMC patterns in relation to toxicity and self-disclosure. The samples are drawn from Youtube.com, Facebook.com and 4Chan.org comments on the controversial advertisement video from Gillette called "The Best Man Can Be". During the analysis, we observed a pattern, where users became more toxic in their communications when the levels of their identification were lower. Moreover, former literature suggests that CMC and social anonymity should increase users' self-disclosure. However, the current research could only partially support such a notion. Instead, we observed that users with a low level of identification (e.g. anonymous) were more transparent in their controversial and toxic attitudes and opinions during brand-related communications, rather than being generally more self-disclosing. This led us to suggest an expanded and updated definition of CMC self-disclosure and user transparency. Whereas self-disclosure focuses on user revelation of private information, such as family status, age, occupation and etc., the user transparency describes revealing deep-seated attitudes, opinions, feelings, and beliefs, that can often be perceived as controversial, toxic or sensitive.
Table of Contents
Introduction 4 Research Question 6 Academic Relevance 7 Practical Relevance 8 Theoretical Framework 9Toxicity of Anonymous CMC: Interpersonal Distance and Disinhibition Effect 10 Self-Disclosure of Anonymous CMC: Hyperpersonal Interactions 14
Methodology 16 Sample 16 Operationalization 20 Codebook 24 Results 26 Toxicity 26 Self-disclosure 30 Discussion 32
Redefining the Self-Disclosure and Transparency 33
Identified Users CMC 36 Anonymous Users CMC 37 Pseudonymous Users CMC 38 Conclusions 39 Study Limitation 41 Future Recommendations 43 Sources 44 Appendix A 49 Appendix B: Memos 70
Introduction
The ongoing digital world has given many people a mouth without a face – the unique ability to anonymously or pseudonymously express their opinions and participate in various online discussions. To put the scale of online user identification in the broader context of social media: around ¼ of Twitters' 326 million monthly users are pseudonymous, meaning that their real identities are unknown. They represent that approximately 89 million people on Twitter are possibly exhibiting altered communications patterns, both positive and negative, that anonymity and pseudonymity might afford them. Moreover, 86% of internet users reported taking steps to hide their identities online (Rainie et al., 2013). This might have significant ramifications for the overall characteristics and climate of social media sites (SMS) environments, computer-mediated communication (CMC) and electronic word-of-mouth (eWOM) circulation (Reichelt, Sievert & Jacob, 2014).
While the term anonymity stems from the Greek word anonymia, which literally means “without a name”, many Web 2.0 consumers adopt certain pseudonyms, called nicknames or usernames, to protect their real identity. Some social media platforms, such as Facebook.com, encourages people to profile their real names, most likely to increase user accountability for their communicative actions (Van der Nagel & Frith, 2015). However, people still choose to hide their real identities. Other SM platforms such as 4Chan.org encourage anonymous user contributions. Meanwhile, most of the Youtube.com users engage in CMC under pseudonymous identities. In the past, YouTube has unsuccessfully encouraged users to adopt real names to increase accountability for toxic comments (Tate, 2012).
But why do people want to be unidentified while engaging in CMC? Kang, Brown and Kiesler’s (2013) research shows that users adopt anonymous (or pseudonymous) identities for various reasons, including engaging in illegal activities and expressing hate speech or other unpopular opinions. Apart from legally/morally dubious reasons, anonymity affordances are valuable to users in some cases, as they allow them to be open in socially delicate or even shunned discussions, such as pornography, drugs, islamophobia, homosexuality, etc. (Leavitt, 2015; Peddinti, Ross & Cappos, 2017; Van der Nagel & Frith 2015). However, the normative and ethical aspects of online anonymity are still debated (Bodle, 2014; Turculet, 2014).
Positive socio-psychological facets of anonymity and computer-mediated communication (CMC) have been a topic of interest for a number of decades now (Kiesler et al., 1984) (Walther, 1996). From a social psychology and communication science perspective, anonymity changes the way people form their identities online and express themselves on the web. Some studies do affirm that anonymous (or pseudonymous) online communication seems to be bringing both functional advantages and disadvantages to the people involved. Anonymous CMC can be either positive and socially acceptable or negative and socially undesirable effects on online communication when compared to real-life or identified interactions (Lapidot-Lefler & Barak, 2012; Walther, 1996). Anonymous CMC seems to allow the amplification of one's social behavior or levels of self-disclosure and engagement. Such positive elements of anonymous CMC usage allow, in some cases, higher levels of self-disclosure (Kiesler et al., 1984; Suler, 2004), lowered social anxiety and desirability (Joinson, 1999) and higher sense of privacy (Bodle, 2014) which leads to more engagement of users in online discussions (Lapidot-Lefler & Barak, 2012). Walther (1996) argued that CMC surpasses face-to-face communication in self-disclosure (Walther, 1996). Other researchers
support such conclusions, having observed that the participants reached higher levels of self-disclosure by using CMC, when compared to real-life conversations (Joinson, 1999, 2001; Kim & Dindia, 2011). Joinson (2001) found that visual anonymity, mediated by lowered public self-awareness, empowered CMC users to self-disclose about their lives even further.
Meanwhile, the negative and undesirable effects of CMC elevate toxicity, griefing, aggressiveness (Joinson, 1999; Suler, 2004; Van der Nagel & Frith, 2015), lack of accountability, invoked digital incivility (Lefler & Barak, 2014), and accelerated rumoring and dissemination of disinformation, distortion of facts and topics (Cho & Kim, 2012). Such toxic effects hinder online communications, making it unproductive and inhospitable environment for beneficial discussions, which, logically, should also hinder CMC users’ ability to self-disclose.
Research Question
The aim of this study is to examine whether different levels of users’ identification (anonymous, pseudonymous, and identified) have an effect on users’ self-disclosure and toxicity patterns. The starting point of the research is to analyze negative brand-related communications of identified users, and to compare them with anonymous and pseudonymous users’ CMC. Current study aims to answer whether anonymous, pseudonymous or identified users display most toxic communications or how do they differ in communicating toxicity. Moreover, which of the groups display highest self-disclosures and if they differ in what type of self-disclosure they communicate? A grounded theory approach is used to examine such trends and come up with new theories regarding toxicity and self-disclosure in brand-related CMC.
RQ:
1. How do users with varying levels of identification (anonymous, pseudonymous & identified) communicate their brand-related CMC with regards to toxicity and self-disclosure?
To better understand the ramifications of varying degrees of identification on user's self-disclosure and toxicity facets, this research will conduct an in-depth qualitative content analysis of users’ comments on Gillette’s new controversial advertisement video “We Believe: The Best Men Can Be" (Gillette, 2019) with grounded theory approach. More specifically, the samples are drawn from Facebook.com (identified), Youtube.com (pseudonymous) and 4Chan.org (anonymous). The comments are analyzed and compared based on varying levels of online identification. Moreover, the current research is not focusing on the specifics of different social media platforms, but rather concentrates on the anonymity and pseudonymity CMC affordances it offers the users based on the affordance theory (Gibson, 1974, 2014).
Academic Relevance
Lack of research makes the current study academically relevant. Anonymous and pseudonymous toxicity and self-disclosing communication characteristics have been overlooked in academic research and brand-related communication context. As a result, the current research aims to bring light to the topic and expand the academic knowledge on anonymous and pseudonymous online consumer behaviors using a grounded theory approach.
Nevertheless, many notable studies (Joinson, 1999, 2001; Pedersen, Razmerita & Colleoni, 2014; Qian & Scott, 2007; Tanis & Postmes, 2007) investigated users' CMC behaviors with online anonymity in mind. However, most failed to consider and explore the implications of anonymity and pseudonymity affordances provides to the consumers in brand-related CMC situations (Schivinski & Dabrowski, 2016).
Lastly, there are contradictory conclusions in the literature on anonymity (and pseudonymity) effects on users' CMC with regard to self-disclosure (Joinson, 1999, 2001; Qian & Scott, 2007; Suler, 2004) and toxicity (Kwak, Blackburn & Han, 2015; Lapidot-Lefler & Barak, 2012). On the one side, researchers suggest that anonymous users are more engaged and self-disclosing, which increases the productivity of CMC. While other studies suggest that users exhibit are more toxic and bigoted behaviors, which would harm the capacity of the overall communication environment. To sum up, different findings suggest, that anonymity can be both harmful and helpful in CMC. The current study's grounded-theory approach re-analysis and testing of such contrasting conclusions of prior studies prove academically relevant, as it aims to explain such contradictory results and offer updated definitions and theory.
Practical Relevance
As the contemporary role of communication professional shifts from a one-way to two-way communication, it is relevant to understand how the audience communicate on SM platforms which provided anonymous or pseudonymous affordances. Unidentified users are an immense chunk of overall interactions on SMS - approximately ¼ (Peddinti et al., 2017), thus, it can have a significant impact on brands’ eWOM circulation (Reichelt et al., 2014). To add to this,
previous studies (Tend, Khong, Gong & Chong, 2014) have shown, that when compared to traditional marketing, eWOM marketing has a bigger overall impact on the brand image and purchase intent. However, it is unclear what effect does anonymity and pseudonymity have on brand-related communication and eWOM in SMS. Consequently, also not much is known how to manage and consider anonymous and pseudonymous brand-related CMC when creating online marketing strategies. By understanding how anonymous and pseudonymous users differ from fully identified users, PR practitioners might be able to take such knowledge into account. It is practically relevant and useful when creating advertisement campaigns, managing brand image online, trying to boost the right type of brand-related engagement and picking the right SM platforms to focus their communication efforts.
Theoretical Framework
Online identity is not a static concept as internet users have different levels of self-disclosure on their profiles. Peddinti et al. (2014; 2017) provide a simple, yet useful methodology for classifying online social media platform user profiles based on their level of identification. Users can be:
1) Fully anonymous – first and last names are absent, no personal user profile picture
is used. no pseudonymous profile names. SM platform, such as 4Chan.org consists of anonymous user' CMC.
2) Pseudonymous (Partially anonymous) – Nicknames used, first or last altered names
can be present, but not both at the same time, profile pictures are usually not personal if used. Youtube.com is an example of a SM platform, which supports pseudonymous user contributions.
3) Identifiable – both first and last names are used. The personal profile pictures are
used by identified users. Facebook.com is a good example of the SM platform, which mainly consists of identified users.
In the sections below, we identify a few major theories, that could explain the contradictory (positive and negative) effects of pseudonymity and anonymity on user' CMC; Interpersonal distance (Argyle & Dean, 1965), hyper-personal interactions (Walther, 1996), disinhibition effect (Suler, 2004), and reduced-cues model (Sproull & Kiesler, 1986) theories can describe different anonymous CMC characteristics and explain how they adversely bring elevated self-disclosure (positive) and toxicity (negative) pathways in users' communication. Moreover, we review the literature surrounding these concepts regarding anonymous, pseudonymous and identified user' CMC.
Toxicity of Anonymous CMC: Interpersonal Distance and Disinhibition
Effect
Researchers and even casual internet users would most likely agree that peoples' communicative behaviors online are more intense and sometimes - even rather toxic when compared with the real-world interactions (Lapidot-Lefler & Barak, 2014; Suler, 2004). CMC paired with anonymity and has also been thought to permit socially undesirable and anti-social aspects of communication, such as aggressiveness, hatefulness (Kiesler et al., 1984) incivility, flaming and other general toxicity (Lapidot-Lefler & Barak, 2015). Because of such detachment from socially acceptable norms of communication, CMC has also been described as less productive and less effective than personal face-to-face communication. The three theories that could explain such toxic behaviors are; Interpersonal Distance (IPD) (Argyle &
Dean, 1965), online disinhibition (Suler, 2004) and (lack of) reputation schemes/stranger on the train effect (Bernstein et al., 2011).
Interpersonal Distance
To begin with, one of the very basic and fundamental features of anonymous and pseudonymous CMC that might be influencing such detachment from socially acceptable manners (toxic) of the communication might be explained by interpersonal distance theory (IPD) (Argyle & Dean, 1965). It is an important factor when considering any type and form of online communication. In very basic terms, IPD states that the actual physical distance between two or more communicators influence how their communication unfolds. In most cases, communication changes occur due to a lack of nonverbal cues and a lack of physical presence. Therefore, in most text-based CMC scenarios, IPD cancels the ability of two or more communicators to gather any non-verbal cues and feedback on each other (Lefler & Barak, 2012). Other studies describe this phenomenon as cues-to-identity, whereas rich communication consists of low IPD (Tanis & Postmes, 2007). CMC discussions are usually text-based, which means that people who engage in casual CMC, such as brand-related discussions, are far-away unknowns to each other and are devoid of any real, proximal social interaction and nonverbal (reduced) cues (Rice & Love, 1987). Eye-contact is a very important non-verbal cue in real-life communication, that dictates interpersonal communication and relationship building aspects such as intimacy, identification, and association (Argyle & Dean, 1965) and is completely vacant in text-based CMC. While emojis could potentially enrich text-based CMC as non-verbal cues (Pavalanathan & Eisenstein, 2015), they are still a rare sight on platforms such as YouTube, Facebook, and Reddit where a lot of text-based discussions take place.
To sum up, CMC cancels verbal cues and increased IPD distances people from each other, alienating them in the process and making it harder for relationship building. In theory, IPD is one of the reasons why that should be causing elevated toxicity in CMC.
Online Disinhibition
Online disinhibition effect theory can also explain some of the toxic facets of unidentified users' online communication (Suler, 2004). It is also largely grounded and builds upon interpersonal distance theory (IPD) (Argyle & Dean, 1965). The empathy deficit is one of the reasons featured in this theory; it states that people online have less identification with each other due to lack of verbal-cues, and as a result - increased IPD. This causes unempathetic feelings towards other users and may lead to cyberbullying and other toxic behavior (Misoch, 2015). Moreover, in real-world discussions (with proximal IPD), people feel the pressure to respond quickly and avoid awkward moments. However, text-based online discussions are still more asynchronous in most cases, allowing people extended timeframes to respond online without losing face. Responses can take anywhere from minutes to weeks in-between, if not evermore (Suler, 2004). As such, lower synchronicity allows users not having to deal with responses immediately and, based on prior theories of IPD and disinhibition, this continues to distance and dissociate people online and strengthen the disinhibition effect and toxicity that comes with it even further (Suler, 2004). Online disinhibition theory essentially argues that people in CMC scenarios are more likely to disregard and neglect other peoples' emotions and feelings while communicating with them online.
Lastly, social and civil norms of communication are probably not as important in anonymous CMC interactions, because due to anonymity, traditional real-life reputation schemes are made impossible (Bernstein et al., 2011). One of such anonymous CMC affordances is “stranger on the train” effect, which states that users tend to be more open, either positively or negatively, if future communication with somebody is highly unlikely (Misoch, 2015). This makes anonymous CMC inherently “unnatural” and different from real-life communication in fundamental ways. This means that users dismiss any possibility of the repercussions of their actions. As a result, people tend to amplify their interaction intensity, meaning that if they hate a product or service, they might be more likely to uncivilly/toxically or very openly voice their opinion on the matter while fully anonymous.
Moreover, any societal or cultural control of communicative norms are obstructed by anonymous CMC. There is also a complete lack of legal or reputational accountability while being fully anonymous. Such anonymity bulwark, that protects the user from the consequences of their actions paired with extreme IPD and online disinhibition effect, might be a huge issue for civil discourse on the internet. This lack of reputational systems and societal control might allow “out of control” users to be extremely toxic and disregard other users’ or topic subjects’ well-being.
Summary of Toxic CMC
All the negative aspects of anonymous CMC affordances lead us to believe that anonymous users will convey more toxicity and incivility when engaging in brand-related CMC. This might be true due to interpersonal distances, lack of empathy, disinhibition effect and most importantly - lack of accountability and reputational systems. Based on prior research, Anonymous brand-related CMC users should also be less afraid to use aggressive, socially
undesirable and even hateful language, because they most feel like there is no aftermath or consequences of their actions since their real identity is hidden by anonymity (or pseudonymity). Thus, we predict to observe higher levels of toxicity in a fully anonymous sample, when compared to the identified and pseudonymous sample.
Self-Disclosure of Anonymous CMC: Hyperpersonal Interactions
Prior research concludes, that CMC users are more self-disclosing (Joinson, 2001; Kim & Dindia, 2011). Similarly, Walther's (1996) CMC hyperpersonal interaction theory suggests, that in some aspects, such as the ability to self-represent, self-disclose and even making new friends, CMC communications can be even more productive than face-to-face interactions. This is due to the fact, that when compared to real interactions, certain CMC platforms afford users better means of constructing their representative and self-images to the audience. Walther's theory of hyperpersonal communication, much like online disinhibition effect theory (Suler, 2004), also relies on a lack of nonverbal cues and subject distance to explain how CMC can lead to higher self-disclosure (Kim & Dindia, 2011; Walther 1996, 2007). Due to lack of synchronicity, immediate and consequential-impactful feedback and nonverbal cues, the anonymous sender is able to strategically compose a goal-oriented message in a self-idolizing way leading to heightened self-disclosure and representation (Kim & Dindia, 2011). Merkle and Richard (2000) speculated that due to anonymity, gender roles play a lesser part in self-disclosure than in face-to-face (or identified) communication since anonymity allowed users to break-free of gender communication constructs. A recent meta-analysis (Kim & Dindia, 2011) of gender roles and self-disclosure using CMC offers similar conclusions. Much like minimization of gender communication roles in anonymous CMC, Van Ruler (2004) noted, that in an anonymous online environment, different subgroups of people (e.g.: sexual,
racial or religious minorities) have a more equal opportunity and, as a consequence, the tendency to express themselves, as minimization of authority and status effect takes place due to anonymity affordances, interpersonal distance and the notion of lack of consequences/shame that would otherwise follow in real life.
The above-mentioned findings suggest that even minorities and other groups that, arguably, hold less power in the "crowd culture" are more self-disclosed in CMC. Can this also mean, that CMC anonymous affordances, also permit other people, regardless of their gender or other sub-group, to more truthful and honest about themselves and their opinions online?
Since prior research suggests that people tend to be more truthful, critical and even hateful while communicating anonymously online, we believe, that this might have a spillover effect in general online discussions and eWOM communications. Since anonymous people are protected by anonymity mask, they are also not afraid of any punishment or any real-life consequences for their honesty and opinions. Moreover, the same applies to self-disclosure, since the real identity is hidden, the user cannot feel any social shame if he/she expresses and disclose intimate opinion/attitude or belief in the context of brand-related communication. Joinson (2001) findings already suggest that anonymous CMC results in higher levels of self-disclosure. Therefore, based on hyper-personal interactions theory (Walther, 1996) and the CMC self-disclosure phenomenon (Joinson, 2001), it is logical to assume, that fully anonymous text-based CMC affords people to be more honest and do not filter their language and opinions as much as fully identified users in eWOM context.
Based on the existing theoretical framework, we predict anonymous users' CMC to be more self-disclosing when compared to identified or pseudonymous brand-related communications. This is due to the hyperpersonal interaction theory, which states that people are able to self-represent themselves to a higher degree. Moreover, as anonymous CMC is less synchronized, high in IPD and having disinhibition inducing effect, we predict users to be more self-disclosing.
Methodology
Design
Exploratory and inductive, the study and research question call for qualitative research tradition. There is no major theory explaining brand-related anonymous CMC, toxicity, and self-disclosure. Qualitative research methods will allow the researcher to gain more in-depth knowledge on the topic, in order to answer the main research question. In addition, the study has a comparative aspect - the identified user-generated content will be analyzed in qualitative tradition and compared with anonymous (and pseudonymous) user-generated responses in 3 different social media. Atlas.ti software is used to code the sample data.
Sample
The sample data are responses to the controversial advertisement campaign from personal care product company Gillette. The video is called “We Believe: The Best Men Can Be” (Gillette, 2019). Currently, with over 30 million views on YouTube.com, Gillette's “We Believe: The Best Men Can Be” is one of the most controversial advertising campaigns of 2019. The video has 30 million views and over 426.000 comments, and is the second-most disliked video of
the year on Youtube.com. Because of the advertisements' controversial political undertones, it attracted a lot of heated and highly engaged user CMC discussions. Moreover, politically fueled brand-related CMC can regularly cause passionate and prolonged discussions and even breed toxicity and invite self-disclosure, which makes it a good sample for the current study. The sample of anonymous users' brand-related CMC is drawn from different social media platforms; the anonymous and pseudonymous users' communications are collected from 4Chan (N=30) (www.4chan.org) discussion board and YouTube (N=33) (www.youtube.com) comments. Meanwhile, identified users’ sample is collected from Facebook (N=70) (www.facebook.com) comment sections. The 3 platforms have a very different user base when it comes to user profiles and their identification. The majority of Facebook's user base is identified. Meanwhile, YouTube’s user base, judging from the preliminary sample of the comment section, seems to be a mix of partially identified users and a mass of pseudonymous users, whose real identity is still hidden. However, the 4chan user base is entirely and thoroughly anonymous, meaning that unlike in YouTube or Facebook, profile pages on 4Chan do not exist allowing full anonymity affordances to the users. This gradient sample of different levels of (non)identification allows the researcher to draw better comparisons between dependent variables (DV: toxicity and self-disclosure) and independent variable (IV: anonymity).
All things considered, the current sample characteristics are fitting and advantageous for the research, because:
1) YouTube, Facebook, and 4Chan comment sections are usually lacking emojis, thus
nonverbal cues that would prohibit online disinhibition effect and hyperpersonal interaction are excluded.
2) Sample of both fully identified and fully anonymous (and pseudonymous) user-generated
content.
3) controversial topic is expected to produce more engaging discussions, which would lead to
richer data.
4) topic subject is revolving around consumer-goods produced by Procter & Gamble
corporation, which fits the research focus on brand-related CMC.
Sample Criteria, Selection, and Limitations
A convenience sampling approach was used to collect users' data. Only comments on the subject of Gillette's "We Believe: The Best Men Can Be” ad campaign met the sample criteria. Moreover, for the comments to meet the criteria required, they have to be of substance, with clearly stated opinions and understandable point-of-view. Since the research is focused on how anonymity and pseudonymity has a relation with toxic online behavior and self-disclosure, only negative attitude comments towards the Gillette’s campaign will be used for the sample. This is because the study specifically aims to examine the trends of how different levels of online identification can influence toxic patterns and self-disclosure, which requires negative attitude expression.
There are disadvantages in the sample. First of all, there is no way for the researcher to verify, if the fully identified user is actually real, or whether he/she has taken on a fake identity with the account (malicious). However, such a phenomenon, presumably, should occupy only a very small portion of the sample - less than 1% (Boshmaf et al., 2015) and is only applicable to fully identified sample group, thus, should be insignificant in the end. Lastly, a lot of sample screening is required. For one, since YouTube, Facebook and 4Chan
are international platforms, with the English language as the most commonplace/default languages to interact between different cultures, there are many user comments with very poor, insufficient English language proficiency or people who still choose to use other languages than English. These comments will be excluded from the final sample coding. Moreover, many users' comments were very short and not suitable for sample.
From the large pool of potential data (46,000 - Facebook.com, 432,905 - Youtube.com and N/A from 4Chan.org) comments were sorted by the newest responses. The main criteria are which include; negative attitudes towards the advertisement, appropriate substance/size. Sorting by "new" gave the most randomized and neutral results when compared to by "most relevant" or "top-comments", because most relevant/top comments tended to have positive attitudes towards the advertisement - undoubtedly, a consequence of social media algorithms, which prioritize positive valence (Lee, 2018). Many of the observed comments (~3000) did not meet the sample criteria. Thus, only N=133 was selected, which means approximately 1 in 20 comments contained the required negative valance, workable structure, substance, and topic. Moreover, there are also other criteria, which is exclusive to the specific sample groups:
1. The sample criteria for fully identified users on Facebook.com are; First and
last name is perceived to be a real name, rather than a username or fictitious pseudonym or character name. Fully identified sample also has to have an apparently real profile picture of themselves in the social media profile.
2. From Youtube.com, we collected the pseudonymous user sample. Collected
only if no real name is disclosed as the username. Moreover, real-life face profile pictures might compromise the user’s pseudonymity sense and, as a result, might influence
communication patterns. Therefore, users with pseudonymous usernames and real-life face user-profile pictures will also be excluded from the sample.
3. Meanwhile, fully anonymous users’ sample was collected from 4Chan.org. The
platform allows fully anonymous contributions; therefore, all data meets the criteria as long as the users' CMC is on the subject matter and of some sort of substance, that could be analyzed.
Operationalization
Justification for creating initial codes for the qualitative content analysis came from a theoretical framework. Moreover, some of the codes are borrowed from previous studies and literature on similar topics of toxicity (Anderson, Yeo, Brossard, Scheufele & Xenos, 2016) and self-disclosure (Joinson & Paine, 2007). However, during test-round open coding (see appendix B), some of the codes from previous studies were reconceptualized and tailored to match the current research. Some of the codes were formulated by observing the phenomena in the natural setting while open coding the sample and seeing the necessity to include them to answer the research question. Overall, the abstract and wide range of behaviors of users engaging in CMC were categorized into 11 different codes spanning toxicity and self-disclosure of personal information
Toxicity
Based on the theoretical framework, we predict that when compared to fully identified users, fully anonymous people display a higher level of toxicity in brand-related discussions on
social media. This means that the researcher has to study and observe toxic behaviors in content analysis. In the scientific literature, flaming, which is synonymous to toxicity was described as ‘‘hostile and aggressive interactions via text-based CMC" (O'Sullivan & Flanagin, 2003, p. 69) or, in parallel with incivility, as something opposite to healthy and productive dialogue (Anderson et al., 2016). Prior research has found that the toxic CMC most commonly manifests itself as direct insults and foul language, mockery and ridicule, argument incitement and threats. All of these were created into codes. During the preliminary test with open-coding rounds, two more uncivil and toxic ways of communication were noticed. One of them is a more subtle way of mockery, use of sarcasm or irony. It was quite a noticeable way to express mockery and dissatisfaction, therefore, it was added to the list of codes. Moreover, harder toxicity was also observed to be present in the open test coding - racism and the use of racial slurs. Therefore, the code for racism and other bigotry was included.
Knowing this, we construct the necessary subsequent codes to observe overall toxicity. The codes are as follows:
1. Foul language - which includes profanity use, not necessarily aimed at other users or topic subject.
2. Mockery - ridiculing, childlike behaviors and other socially unacceptable ways of communication.
3. Attacks - malicious insults targeted at other users or topic subject
4. Threat- intimidation targeted at other users or topic subject. Users expressing the conclusion that he/she will do something to hurt the subject of topic.
5. Distortion - perceived exaggeration of facts or topics, clear bias towards misinterpretation and misrepresentation of information on the topic.
6. Sarcasm/Irony - use of sarcasm in a mocking remark to express underlying dissatisfaction towards the subject.
7. Bigotry- racist remarks, racial slurs, offensive/toxic generalization and prejudice and discrimination against a certain group of people.
Lastly, a code, which marked a civil discourse was created, because, during open coding sessions, there were comments that did express their dislike to the advertisement, however, these users’ messages did not contain any toxicity. This code was necessary to highlight the contrasting trends of toxicity and civil discourse between the different sample groups.
8. Civil Discourse - use of politically correct, socially acceptable ways of communication to express the dislike to Gillette’s advertisement. User CMC that uses Civil Discourse usually is void of any toxicity.
Self-Disclosure
We predict that when compared to fully identified users, a fully anonymous user on social media will display higher levels of self-disclosure in brand-related discussions. Considering the sample characteristics, self-disclosure was a particularly hard concept to measure and categorize qualitatively. Nevertheless, simple codes were created to see patterns of higher, lower or non-existent levels of self-disclosure in different sample groups. As Joinson and Paine (2007) pointed out, not all self-disclosure is equal, meaning that disclosing "... not another dime for Gillette" is not the same (higher) level of disclosure as "I had been using your products since I was 15, I'm 61 now. just threw out what I had of them, and will not be using anything...”. Two overarching codes were constructed to survey real self-disclosure
(major self-disclosure) and self-representation (minor self-disclosure) levels in the content analysis. Moreover, a code was also included to consider and highlight the lack of any self-disclosure in some cases (Self-preservation). From this, we were able to conceive the following codes on different levels of self-disclosure:
1. Minor disclosure - superficial statements, a less intimate/private concession of self-information that would otherwise be unknown to others (Kim & Dindia, 2011). Prior studies also found the importance of small cues when using CMC to self-represent (Ellison, Heino & Gibbs, 2006). Small cues can be seen as minor disclosures for self-representation purposes - hints of personal facts from real life, but not concrete personal information (Ellison, Heino & Gibbs, 2006). It can also be subtle cues, with no concrete information about the user.
2. Major disclosure - intimate and personal statements- a higher level of self-information revelation. It can include intense, profound and concrete private information. Major disclosures are clearly expressed statements based on users' real-life facts, that bypass subtle cues. In short, major CMC disclosures confer more about the users' real-life than minor disclosures.
3. Self-preservation - users involvement in the discussion without any self-disclosure or self-input. Such a user might still choose to engage and provide arguments. However, the only self-disclosure that can be visible is the (often vague) personal stance/opinion of the commentator. Users who self-preserve do not give out any personal information and past or future purchase intents.
Lastly, minor and major self-disclosure codes do not cover the users’ communications on more abstract aspects, such as emotions, attitudes, feelings, etc.
Codebook
Below is the code table of the qualitative content analysis, with the quotes and labels of sources from the actual sample. Codes from 1-7 are for examining toxicity, meanwhile, codes 8-10 are for self-disclosure:
Table 1. Codebook: toxicity Toxicity
N
o. Code Example Quote Label
1. Foul Language
“holy shit the gillet ad was fucking gay. fuck those cunts…”
Anonymous 4Chan Case 5
2. Mockery
“HAHAHA the marketing director was too afraid to say "no" to a feminist VP's ad pitch. So this is the consequence.”...
Imashaaark YouTube Case 24
3. Attack
“@ Gillette .., go f yourselves .., you anti male fem libtard a- holes, is that all the credit you can give towards men .”...
David Levy Facebook Case 17
4. Warning
“we REALLY need to do is boycott Procter and Gamble.”...
PAN DA YouTube Case 31
5. Distortion
“This is the best advertisement for growing a beard that I've ever seen.”...
ALG Bass YouTube Case 29
6. Sarcasm/Iro ny
“My wife's boyfriend says this will be the best commercial in 2019’’...
Guybrush Threepwood YouTube Case 26
7. Bigotry
“All of you Africa Americans are responsible for the behavior of a few niggers. Stop being niggers". Stop forcing sjw horseshit on the American”...
Anonymous ID:qBeBe7f L 4Chan Case 21 8. Civil Discourse
‘’What a stigmatizing commercial, so full of political correctness!!! No more Gillette for me because I'm not a predator, just a normal man that lives a normal life!!’’
Brynjulf Ei Facebook Case 44
Table 2. Codebook: Self-disclosure Self-Disclosure
N
o. Code
Explanation and Quote Label
9. Minor disclosure
“Just used my last disposable... not another dime for gillette”... Sean Brooks Facebook Case 57 10 . Major disclosure
“ For my own small business, I pray on hand and knees that my marketing department gives me some indication that they're about to release (...)
I thank you Gillette for demonstrating to me why I should keep a tighter leash on my marketing department.”...
Michael L. Wells III Facebook Case 65 11 . Self-preservatio n
“Next time make sure your advertising company isn't full of Marxists that hate your customer base.”..
Anon None YouTube Case 20
Results
Judging from the coding, we could see clear divergent trends of toxicity and self-disclosure in different sample groups from 4Chan.org (anonymous), YouTube.Com (pseudonymous) and Facebook.com (identified) (see table 3 for a summary).
Toxicity
Identified User CMC
Uncivil and toxic CMC was observed in all the groups. However, as predicted, most civilly discourse CMC was carried out by fully identified users. Identified users tended to use the least foul language and bigotry. Instead of toxicity, they most notably civilly voiced their strong discontent towards Gillette's new advertisement, mainly by using minor self-disclosures and on future purchase intent - something that could be described - "consumer power":
“Despite using your product for the past 35+ years I'll be trashing my remaining razors and using a product from another company”. - John Killpack, case 1
Pseudonymous User' CMC
YouTube.com pseudonymous users, while being less toxic than anonymous users, were still considerably more uncivil in their discussions when compared to identified users. Mockery sarcasm and distortion of facts and topics were eminent in pseudonymous CMC. One user mockingly and sarcastically wrote:
“Whole my life when my father told me that im a man he was telling me that im a raping toxic human.Thank you for getting me out of the dark Gillette.”
- Kenan, case 6.
Presumably, this pseudonymous CMC includes many toxic elements such as mockery, sarcasm, irony, and distortion of the topic. While no foul language is used, this specific user did not hesitate to use harsh words such as “raping toxic human” in order to mockingly and sarcastically describe how they were depicted in the advertisement and, in turn, voice his discontent towards the Gillette’s advertisement campaign. Such eccentric and borderline toxic behaviors masked by mocking/comedic attempts were commonplace among pseudonymous CMC.
On a side note, YouTube’s pseudonymous users were observed to be politically motivated more than identified users:
“HAHAHA the marketing director was too afraid to say "no" to a feminist VP's ad pitch. So this is the consequence.”
- Imashaaark, case 24.
“...Gillete has fallen to Feminism…” - Half-Life-Lore, case 5.
“Next time make sure your advertising company isn't full of Marxists that hate your customer base.”
- Anon none, case 20.
- E F, case 22.
Attacking underlying ideological messages of the advertisement were more common than disclosing lowered levels of future purchase intent, as seen in fully identified users.
Anonymous User CMC
Lastly, the fully anonymous group was observed to be undoubtedly the most toxic out of the three groups. Civil discourse was a very rare sight amongst anonymous users. Moreover, most of the posts contained hard profanity and even bigotry, such as racism and sexism. Such intolerant toxicity in most cases was noticeably also followed by poor spelling and uncalculated syntax:
“fuck typical faggot response. the gillett add doesnt distinguish between good men and shit cunts, ergo it implys that all men are shit cunts.”...
- Anonymous, case 2.
Moreover, the use of toxic distortion of facts and topics was examined to be far-reaching amongst anonymous users. Some of the users adeptly misrepresented the advertisement and even Gillette's management team' "hidden" motives:
“it's probably just a tumblrette scourge sneaking into the company and subverting it. tumblrette marketing scum: let's make a man-hating ad!”
- Anonymous, case 28.
When compared to identified and pseudonymous users, such as distortion and dissemination of misinformation, and "conspiracy theories", was a noticeable trend in the current sample group of anonymous CMC:
“but lets be absolutely clear about one thing. this gillett ad isnt about making men better. its about riding the wave of anti-male sentiment sweeping across”
- Anonymous, case 2.
One interesting finding is that the use of sarcasm and mockery was considerably less noticeable in anonymous CMC. While identified and pseudonymous users used that to communicate the discontent, the anonymous group tended to use bigotry, foul language, and almost comical distorted CMC to voice their distaste for the Gillette's new campaign.
Overall, the analysis of the three groups of different levels of identification (identified, pseudonymous and anonymous), revealed that the fully anonymous people tended to display the highest level of toxicity in brand-related CMC. While identified and pseudonymous users seemed to use more sarcasm/irony and socially undesirable mockery to unleash their toxicity, anonymous users were overall bigoted, racist and distorted the facts and topics on the subject, meanwhile using foulest language out of all the groups.
Pseudonymous user comments seemed to contain more toxic trends, compared to identified users. They used mockery and foul language considerably more distinctly. However, identified facebook.com users were more prone to use toxic intimidation, masked as consumerism power, against Gillette’s wellbeing as a company. Threats of boycotts, trashing of the products and foul word-of-mouth dissemination were made quite prominently among identified users.
Anonymous group users display a non-proportional amount of toxicity compared to the two other groups. To conclude, anonymity led to more toxic communication in brand-related CMC. Therefore, our predictions that anonymous people display a higher level of
toxicity in brand-related CMC, when compared with identified users were proven correct based on the current sample.
Self-disclosure
To begin with, the most common use of self-disclosure across different platforms was expressed by disclosing their past, present and future actions and purchase intent. Such CMC self-disclosure was above all, noticeable in the identified sample group. Nevertheless, the overall perceived levels of self-disclosure of private information in the sample were smaller than expected in pseudonymous and anonymous groups. Meanwhile, counter-intuitively, identified facebook.com users tended to communicate minor self-disclosures quite often - most of the people reported their past and future attitudes towards Gillette and disclosed lack of future purchase intent because of the advertisement:
“I will be avoiding all of your products from now on.”. - Kyle Rook, case 4
Facebook.com user CMC was the only platform where notable major self-disclosure CMC took place:
... I commit not to purchase any Gillette product ever. Tomorrow, Head and Shoulder (my shampoo for the past 20 years..), the day after pampers (I have 1 baby at home and 1 on the way), etc. ...”
- Yann Le Barz, case 29
In such fashion, identified users’ CMC was observed to be more self-centered and self-disclosing with regards to private information, when compared to pseudonymous or
anonymous users. The two latter groups were more prone to self-preserve their personal details and tended to only express their negative attitudes towards the advertisement.
Worth noting, that while in some instances of minor self-disclosure were present in anonymous and pseudonymous groups, however, no major self-disclosures were observed. The anonymity and even pseudonymity seemed to distance and inhibit the users from inputting their own experiences and other personal information apart from attitudes and opinions into the CMC surrounding Gillette’s advertisement.
The counter-intuitive results led to unpredicted results. Based on the theoretical background on pseudonymous and anonymous CMC, we were expecting to find that anonymity allows more self-disclosure of sensitive-personal information. However, the findings were opposite to that, since the sample of fully identified users tended to minor and major self-disclose considerably more intensively, than pseudonymous and anonymous users in brand-related CMC context.
Table 3. Summary of results
Toxic CMC enactment Self-disclosure Identified Voice of consumer - threats
of boycott, intentions of warning others not to buy, never to buy again themselves, and intentions of discarding the currently owned products.
Strong tendency for minor self-disclosure,
and bit of major self-disclosures.
Pseudonymous Use of mockery and jokes to put the subject topic under a bad light. To prove a point, many sarcastic statements were made.
Weak tendency for self-disclosure. No inclination towards major disclosures.
Anonymous Bigotry, racism and sexism. Distortion tactics were used on the topic subject to make the advertisement and Gillette look even worse.
Non-existent self-disclosure propensity. Most of the communication is with preservation of concrete
self-information. High transparency in controversial attitudes/opinions and feelings
Discussion
The current study aimed to answer the research question:
- How do users with varying levels of identification (anonymous, pseudonymous & identified) communicate their brand-related CMC with regards to the use of toxicity and self-disclosure?
For one, the content analysis results showed that social media users do have different communication patterns based on their level of identification when engaging in brand-related
CMC with regards to toxicity and self-disclosure. The most important findings were that the anonymous sample group tended to communicate more toxicity, meanwhile, the identified group was more self-disclosing and civil. Additionally, we have noticed a trend in the sample, where the more anonymous the platform is, the more uncivil and toxic the negative brand-related user's CMC seems to become. This is evident when comparing the identified group with the anonymous group. Anonymous users were noticeably more bigoted and prone to use socially undesirable ways to voice discontent for Gillette's advertisement. Distortion of facts on the topic, racism, and extremely foul language was commonplace.
Moreover, we propose an expanded definition of self-disclosure with regards to CMC, which would include the new concept of (self-)transparency. Like self-disclosure, transparency is a type of voluntary admission of self-information. However, it is specifically regarding underlying, sensitive attitudes and private beliefs, rather than disclosure of private details and "concrete" information.
Redefining the Self-Disclosure and Transparency of Anonymous and
Pseudonymous User CMC
The findings of various prior studies on CMC self-disclosure and anonymity seem to contradict the findings of the current research. Many studies point out, that anonymous people are more prone to be self-disclosing on their private lives when compared to identified users (Bodle, 2014; Joinson, 1999, 2001; Kiesler et al., 1986; Suler, 2004; Walther, 1996;). Yet, the current study and prior similar studies (Qian & Scott, 2007) could not fully support such conclusions. Opposite effects on anonymity have been observed in the sample - users who engage in anonymous CMC seemed to preserve their privacy and generally emit any of their
private information from their communications. We conclude that anonymous users are not prone to self-disclose in the current definition of the concept. As a result, this research can only partially agree with the prior research conclusions on the topic, that anonymity allows users to become more self-disclosing in various CMC scenarios. However, redefining the concept of self-disclosure might explain why the findings of previous studies could not be supported in the current research.
Previous CMC and self-disclosure studies, (Joinson 1999, 2001; Qian & Scott, 2007) adopted the definition of self-disclosure vaguely and very broadly. The current definition of self-disclosure describes an event where a person communicates any piece of information otherwise unknown to the receiver, including giving "concrete" private information, attitudes, opinions, feelings and etc. However, in the current research, narrowed down and split the broad definition of self-disclosure. We propose altering and dividing the concept of users willingly sharing information to other people into two interrelated concept definitions: self-disclosure and transparency:
1) CMC transparency describes the instances where users are communicating private
and/or sensitive personal attitudes and opinions without the necessity of sharing private "concrete" information on their real-life or past, present and future intents in real-life. CMC transparency can include conferring and being transparent on political, sexual or any other sensitive attitudes, opinions, and beliefs. The qualitative data analysis suggests that fully anonymous users can be more transparent, rather than self-disclosing about their culturally unpopular, sensitive and toxic attitudes, such as expression of racism, homophobia, and sexism. We speculate that the anonymous
CMC transparency phenomena can also extend to the subjects of sexuality, politics, religion and other sensitive topics.
2) CMC self-disclosure suggested redefined and extended concept describes the cases
where users are not necessarily transparent on their attitudes, inner beliefs, and opinions, but rather self-disclose the more "concrete" private information based on their real-life. These can include non-controversial self-revelations, past, present or future intentions, their family status, occupations and etc. The current research concludes that fully identified users tend to self-disclose during brand-related CMC to a higher degree when compared to anonymous and pseudonymous users.
To illustrate why the distinction between the definitions of anonymous CMC transparency and self-disclosure are needed, we could make a hypothetical example of a person whose unpopular or sensitive opinion is silenced by current culture norms, dominant coalitions or in-groups. Such a person might not be able to be transparent in his/her opinion openly without risking social exclusion (Sidanius & Pratto, 1993). The person might be able to use CMC to self-disclose his concrete and explicit personal details; however, he cannot be fully transparent about his sensitive and private beliefs and attitudes while being identified. Therefore, his honest and transparent-self, in such a case, can never be expressed by identified CMC. Proposed theory suggests that distancing/disinhibiting tool, such as anonymous CMC, can help such a person to escape the social dominance and be transparent about his/her personal opinions and attitudes without the risk of reputational loss. Anonymous sample analysis in the current research seemed to point towards the accuracy of such assumptions, that transparency and self-disclosure are two different concepts and definitions used to describe CMC in various CMC scenarios.
To sum up, anonymity did not permit users to be more self-disclosing about their personal information in brand-related CMC, yet it did allow them to be more transparent, or in simple terms - more honest with regards to their dark and most intimate beliefs and attitudes. In the sample, such a theory is supported by anonymous users' CMC trends, which pointed towards self-revelation of their culturally unpopular, bigoted and other socially unacceptable opinion tellings, which cannot be seen in identified group samples. By redefining the definitions of anonymous and pseudonymous information sharing in CMC (self-disclosure vs. transparency), the current study can also partially support the conclusions of previous studies, which suggested that anonymous CMC users are generally more willing to share private information. However, we still maintain that CMC anonymity does not reinforce anonymous or pseudonymous users to be more self-disclosing by our newly suggested definition.
Identified Users CMC
Contrary to our predictions, it was identified users, and not anonymous users, that displayed the highest leaning towards self-disclosure of private information in the comments regarding Gillette's new campaign. On the one hand, the result could be explained by the fact, that primarily identified user based social media platforms, such as Facebook, are designed for people to disclose and share their private information with ease. However, a prior study that examined users' self-reported information self-disclosure motivations found that users are motivated to self-disclose, because it helps them maintain relationships/image and increases platform usage gratification (Krasnova, Spiekermann, Koroleva & Hildebrand, 2010), much like reputation systems (Krasnova et al., 2010). Meanwhile, fully anonymous users, by definition, lack such ambition or even ways to do it, even if they wanted to.
Anonymous Users CMC
Based on the results, anonymous users' CMC did not have a higher degree of self-disclosure in the sample. On the contrary, anonymous users tended not to disclose personal information. Identified users tended to disclose private information to maintain reputation and relationships (Krasnova et al., 2010). However, maintaining image and relationships is arguably harder with pseudonymous account and probably not possible at all with fully anonymous CMC. Not having reputation to maintain eliminates the need for anonymous users to self-disclose for that purpose, which would explain the lack of self-disclosure of anonymous users in the results.
Moreover, anonymous CMC users did express bigoted and socially/culturally shunned opinions more often than pseudonymous and identified users. Expressing such private and troublesome opinions and attitudes while being fully identified or in real-life can even lead to legal troubles. However, while completely anonymous, users experience a disinhibition effect (Lapidot-Lefler & Barak, 2012) and they cannot be socially or legally prosecuted for expressing their unpopular opinions. The sample showed that anonymous people were more highly prone to be transparent about their toxic attitudes. This means that anonymous CMC does not afford users to be self-disclosing on their private information (minor or major), but rather enables them to be more transparent about their true beliefs and opinions, no matter how toxic or unpopular. Transparency of this degree, as was observed in the anonymous sample, would never be viable in a fully identified sample, because only anonymity can afford the users' lack of consequences for their actions and reputational systems. Anonymity also enabled the disinhibition effect and boosted IPD between other users and the Gillette company. All these factors contributed to anonymous user transparency, which translated itself into toxic sexism, bigotry, mockery, and other flaming CMC.
Notion of anonymity as an enabling factor for transparent CMC, rather than self-disclosing CMC can be furtherly supported by the idea of asynchronicity as one of the reasons for the disinhibition effect (Suler, 2004). Asynchronous CMC causes more disinhibition. We suspect fully anonymous CMC SMS without profiles, such as 4chan.org, to be highly asynchronous when compared to Facebook.com and Youtube.com samples. This is because identified or pseudonymous SM platforms have profile pages and notifications. Profiles usually allow users to receive notifications on their CMC discussions, which boost synchronicity of the CMC. Meanwhile, fully anonymous, profile-less SM platforms do not provide such affordance to the users. Furtherly disinhibited people can be more transparent by discharging their personal attitudes/opinions.
Additionally, the lack of online reputation schemes (Bernstein et al., 2011) is suspected to have had a positive effect on the results of toxicity and transparency of fully anonymous users' CMC groups. Meanwhile, the presence of online reputation schemes is presumably boosting self-disclosures and the civil discourse of identified and pseudonymous users. Most of the social media, such as Facebook, Twitter, Instagram, YouTube, Reddit and many more, have some sort of CMC reputation schemes implemented in the platform, such as likes, retweets, upvotes, favorites and ect. These schemes are known to reduce toxic behavior and disinhibition effect (Bernstein et al., 2011) and based on the results, we conclude that it had a debilitating effect on user' CMC transparency.
Pseudonymous Users CMC
When it comes to toxic CMC of the pseudonymous users, the group was leaning closer to fully identified users with regards to toxicity and self-disclosure. This is because there were
little bigotry and foul language. However, they tended to use more mockery and sarcasm as tools to express the dislike for the advertisement. It is worth noting, that based on the findings, even pseudonymous identity seemed to be anchoring and hinder users' ability to be fully transparent as fully anonymous users. We suggest, that ownership of an accouCnt or a profile, even pseudonymous, activates the cognitive systems regarding reputation and negates, to some extent, the "stranger on the train effect" (Bernstein et al., 2011), thus it prevents people from going "full toxic" in their CMC, as seen in the anonymous responses to the Gillette's ad. Lastly, contrary to the identified group, the pseudonymous users tended to be more concealing of their personal information, as self-disclosure instances were minor and few. Pseudonymity, unlike anonymity, allows users to create their own virtual identities, that do not mirror the real-life identities of the users' (Lin & Wang, 2014). This may allow pseudonymous users to detach from their real-life selves and their real experiences, which prevents them to be self-disclosing of their personal information. This would explain the lack of self-self-disclosing trends in the pseudonymous group.
Conclusions
Regarding toxicity, the content analysis results suggested that lower levels of identification, indeed, eliminate accountability of user' communication, as well as distances (IPD) and disinhibits users from using civil discourse techniques to voice their negative opinions on the advertisement. As a result, anonymous and pseudonymous communications in the sample have been observed to be more toxic. While identified users are bound by reputational systems and consequences of their actions, thus, real names restrict higher levels of disinhibition. This led to a more grounded and controlled civil discourse in the identified group.
Moreover, user communications also suggest, that fully identified people tend to self-disclose private information to a higher degree. It could mean, that users with real names tend to be more self-centered, as the high level of identification confines users to their real-life selves and limits the performance of self-constructed “online identity” that might be seen in pseudonymous or anonymous users. This causes identified users’ negative comments on Gillette’s advertisement to be grounded and based on real-life experiences, such as - “I’ve been using Gillette for over 20 years, but not anymore”. However, the current research suggests, that while anonymous and pseudonymous groups showed low self-disclosure tendencies, the latter groups displayed higher levels of transparency into their real and ungrounded attitudes, because they were not restricted with IPD (Argyle & Dean, 1965) and reputational systems (Bernstein et al., 2011) as much as identified group would have.
The current conclusions based on the sample have some practical applications for PR practitioners and social media platform policies. However, the results can be generalized with caution and to a limited extent:
1) Negative eWOM is known to have an impact on brand image and lower
purchase intent (Reichelt et al., 2014). Companies and PR professionals are advised carefully to select the platforms in which to release the possibly controversial content, such as Gillette's advertisement. Sample analysis indicates that fully identified, or at least monitored pseudonymous user platforms minimize the likeliness of toxic CMC/ eWOM that can result in negative affiliations, which might hurt the brand image by introducing unwanted associations (Cantallops & Salvi, 2014). However, this will not necessarily reduce the negative attitudes, just reduce the toxic and uncivil
discourse, that might possibly carry a higher eWOM impact on passive readers (Teng, Wei Khong, Wei Goh, & Yee Loong Chong, 2014).
2) Anonymous and Identified user bases have their own advantages and
disadvantages. CMC platforms should first consider their primary mission, before adopting and enforcing either level of identification for their users. Evidently, anonymous/pseudonymous user bases will be more efficient at being transparent and true towards CMC surrounding sensitive topics and personal attitudes. However, it can also bring unwanted toxicity due to lack of accountability and disinhibition effect. Anonymity and pseudonymity user base can work best on CMC platforms focused on sensitive topics, such as sexuality, religion and etc. However, to reduce the risk of the platform becoming toxic, some monitoring/control precautions are to be implemented.
Meanwhile, identified user CMC platforms are less toxic and more accountable based on the sample data and conclusions of prior research papers. This means that platforms with consumer reviews would be more credible and less toxic if the users are fully identified. In general, identified user platforms, such as Facebook, would produce a more civil environment for user CMC and would require less monitoring to curb toxicity and possibly even lower the negative impact of eWOM (Sen & Lerman, 2007). However, such platforms would possibly diminish user transparency on more sensitive topics due to a lack of disinhibiting effects.
Study Limitation
While the sample results from three different groups of different levels of user identification seem to point to clear CMC pattern differences, the sample size was
