The Bad Boys and Girls of Cyberspace

Abstract

A significant feature of virtual interactions involve online deviance and trolling; these include behaviors that range from mild mischief, to offensive language, to hacking and trolling, and to expression of complex social problem, such as by revolutionaries, freedom fighters, or pedophiles. Yet little research has examined online trolling in general or the impact of gender and context on these behaviors in particular. By focusing on the effects of gender and context on perceptions of and reactions to online trolling, this article enhances Suler and Phillips’ framework for online deviance. Results indicate that men and women react differently to online trolling, and their perceptions of the impact of trolling on online communities vary; men and women identify different motivations for similar behaviors in different communities, and they both perceive that men and women trolls differ in their behavior and motivation. The study concludes with suggestions for future research.

Keywords

online communities behavior gender perceptions trolling reactions context

Introduction

Online trolling is a specific example of deviant and antisocial online behavior in which the deviant user acts provocatively and outside of normative expectations within a particular community; trolls seek to elicit responses from the community and act repeatedly and intentionally to cause disruption or trigger conflict among community members (e.g., Hardaker, 2010; Shachaf & Hara, 2010). Trolling has always occurred online and some early online communities were even comprised entirely of deviants (Jordan & Taylor, 1998; Krappitz, 2012). Research on these online behaviors has focused, for example, on deviants’ motivations (e.g., Jordan & Taylor, 1998; Turgeman-Goldschmidt, 2005), the role of the Internet in encouraging deviance (e.g., Turgeman-Goldschmidt, 2005), enabling factors online (e.g., Denegri-Knott & Taylor, 2005), and perceptions of these behaviors (e.g., Utz, 2005). However, online trolling, in general, and the role of gender in perceiving, motivating, defining, enabling, or reacting to trolling, in particular, need to be better understood. This study aims to focus on how gender impacts online trolling and to understand if men and women perceive and react differently to trolls and if trolls’ gender impact perception of and reaction to trolls and their motives. Likewise, the role of context in the perception of deviance and online trolling needs further scholarly attention. This study also aims to understand the impacts of context on online trolling and to examine if online trolling is perceived differently in different online communities.

As women are becoming more active online (Lu, Lin, Hsiao, & Cheng, 2010), it is increasingly important to understand and analyze the role that gender plays with respect to online trolling; yet, scholarly attention toward trolling in online communities has largely focused on men, likely because male deviants have historically been more numerous and accessible (Suler & Phillips, 1998). Studies in computer-mediated communication suggest that men tend to initiate and value conflict in discourse, while women avoid it (Herring, 2008). Yet it is unclear what role gender plays in perception of the violation or enforcement of norms, though there is evidence that norm construction includes gendered aspects (e.g., McLaughlin & Vitak, 2012).

Research has shown that behavior, both normative and deviant, has been affected by the particular contexts in which they occur online (e.g., Williams, 2000) and off-line (e.g., Mahoney & Stattin, 2000); yet antisocial behaviors in online environments have received considerably less scholarly consideration (Williams, 2000). Context is highly important in shaping sociotechnical interactions and must be taken into consideration when evaluating group outcomes and norms (e.g., Kling & Star, 1997). Specifically, social informatics has emphasized how much context matters in technologically mediated social networks (e.g., Kling & Iacono, 1989; Sanfilippo & Fichman, 2014). Based on preliminary research into trolls’ motivations (e.g., Downing, 2009; Shachaf & Hara, 2010) and perception of deviance (e.g., Utz, 2005), there is evidence that context impacts nonnormative behavior (e.g., Denegri-Knott & Taylor, 2005; Downing, 2009); yet the specific nature of this relationship is unclear. Furthermore, while there is evidence that similar motives shape different behaviors in off-line contexts, there is no comparable empirical evidence for online contexts, and online norms sometimes differ from off-line norms (Maratea & Kavanaugh, 2012). The lack of scholarly attention to trolling is problematic because of the unique contextual attributes of online communities (Maratea & Kavanaugh, 2012), such as anonymity and disinhibition (e.g., Suler, 2004), as well as increased online community awareness of and attention to deviance (Barak, 2005).

Thus, in an attempt to understand the impact of gender and context on online trolling, this article aims to answer three research questions. First, the article focuses on gender differences in perceiving men’s and women’s trolling online and specifically poses the following two research questions: Does gender affect perceptions of men’s and women’s trolling behavior, motivations, and impact on the community? Does gender affect reactions to men’s and women’s trolling? This article then presents analysis of the relationship between context and social understanding of trolling motivation in online communities, addressing a third research question to what extent does context impact perception of motivations for trolling online?

Background

As research is just beginning to deconstruct online deviance, theory within this area is scant. Suler and Phillips (1998) provide a notable and rare typology of online deviant behavior, which also suggests strategies to combat online deviance. This typology of online deviance informs our study on the basis that theoretically situates behaviors like trolling within a spectrum of severity and in relationship to motivations, we seek to extend their work by illustrating and explaining the gendered and contextual dimensions of deviant behaviors.

Suler and Phillips’ (1998) typology is useful because it characterizes many forms of deviance and articulates social and technical responses to online deviance. The various types of deviance in their typology include (1) mild deviance, such as “clueless newbies” and mischievous users such as “parodists,” on “the lower end of deviance” (p. 278); (2) offensive avatars; (3) offensive language; (4) complex social problem expressions, such as by revolutionaries, freedom fighters, or pedophiles; and (5) technocrime (hacking), which includes software exploitation and breaches of restricted databases, on the more serious end of a spectrum of deviance. Yet it is also clear that not all instances of offensive language, for example, reflect the same motivations or implications, as deviance can be either “acting out,” without understanding their behavior, or “working through,” which reflects rational albeit antinormative cognitive underpinnings (Suler & Phillips, 1998). With respect to each behavioral type and expression, social, cultural, technical, and contextual factors are considered, so as to best define corresponding intervention strategies.

Responses to deviance are characterized both in social and in technical terms (Suler & Phillips, 1998), emphasizing the need for engagement rather than reciprocal confrontation. The framework differentiates between preventative and remedial interventions. Preventative measures include identification of clear public standards for behavior, restricted areas bounded by community infrastructure, and thresholds for participation (Suler & Phillips, 1998). Remedial interventions are more numerous and can be categorized as interpersonal, such as polite private discussions, conversions, and “whispering” confrontations to allow trolls to save face; technological, such as banning or disconnecting; fully automated responses, such as using scripts to identify banned words or filters to send users of offensive language to gated spaces separate from the rest of the community; and governance through supervision and policing by overseers (Suler & Phillips, 1998).

However, the framework is notably limited in assuming that deviant behaviors are limited to male perpetrators and to male-dominated online culture, and in that it conceptualizes behaviors and responses without exploring the best responses to specific behaviors or perceived behaviors in comparison to deviants’ intents. Furthermore, the framework ignores context, but the context of online deviance can determine the nature of aggressive and deviant behaviors, as there is evidence that the relationship between context and social understandings of trolling in community interactions may be more complex.

Suler and Phillips (1998, p. 276) acknowledge the potential critical impact of context and wrote that “Every culture and subculture has its own standards about what is acceptable and unacceptable behavior,” yet there is a need to map specific behavioral types as deviant for particular communities; it is unclear, in their framework, if and how context impacts trolling behaviors and effective responses to online trolling.

Further, Suler and Phillips’s (1998, p. 275) article “bad boys in cyberspace,” paid little attention to gender in their work as they acknowledged “there are bad girls in cyberspaces, too, but they do seem to be outnumbered by the males,” while at the same time their argument ignores the role of gender in the discussion of online deviance. However, given their simplistic dismissal of gender as a variable in deviance and the ever-increasing participation of women in online communities of all natures, this appears to be an insufficient explanation of the gendered nature of online deviance, and the need to enhance Suler and Phillip’s framework to account for gender impacts is evident.

Gender

Differences between male and female actors can be construed as either sex or gender; for the purposes of this study, gender is explored on the basis that gender norms and a gendered environment, as opposed to physiological differences, lead to behavioral differences. Gender, in the normative sense, refers to the expected behaviors of men and women as well as their relative positions of power (Lester, 2011). Additionally, gender is socially constructed in such a way as to control people and behaviors in ways that are sometimes counterproductive (Adam, 2009; Puente & Jimenez, 2011).

Research has addressed the role of gender in affecting online norms (Puente & Jimenez, 2011) as well as the translation of off-line norms into the virtual spaces produced by gendered cultures (Adam, 2009; Herring, 2008; Puente, 2011). Female trolls and deviants have not been well studied because they are a minority of online deviants (Adam, 2009; Jordan & Taylor, 1998) and traditionally excluded from online deviant subcultures (Jordan & Taylor, 1998). However, popular press accounts (e.g., Muhammad, 2000) and an emerging body of cyberfeminism research (Puente, 2011; Puente & Jimenez, 2011) document the developing trend of women deviants employing the Internet to accomplish ideological goals.

There is evidence that off-line gendered norms are reinforced online (Barak, 2005; Danet, 2006; Herring, 2008) and that online norms are sometimes more flexible (Maratea & Kavanaugh, 2012; Shachaf & Hara, 2010; Utz, 2005). Yet variables that reinforce gendered norms or lead to gendered equality are unclear, as most scholarly attention toward gender and online deviance has focused on women’s sociopolitical goals (e.g., Adam, 2009; Puente & Jimenez, 2011).

Individuals and communities have a wide range of perspectives about online deviance, from the opinion that these are explicitly equivalent to indiscretion or crime off-line (e.g., Baker, 2001) or that certain behaviors are unacceptable (e.g., Danet, 2006; Whitty, 2005) to perceptions that these behaviors are rationally motivated (e.g., Jordan & Taylor, 1998; Kirman, Linehan, & Lawson, 2012) and even nonproblematic due to their virtual nature (Whitty, 2005).

Research has only begun to address the factors affecting responses to and perceptions of trolling, despite its frequency in online communities (Shachaf & Hara, 2010; Whitty, 2005). It is necessary to better understand how individuals and communities perceive specific acts of trolling; what factors, including gender, impact specific perceptions of online trolling; and how these perceptions relate to specific responses. Focusing on gender, we hypothesized that perceptions of and reactions to trolling would vary by gender of the individual (Hypothesis 1). We further hypothesized that individuals will perceive the same trolling exhibited by men and women differently and therefore reaction to online trolling will also vary by deviants’ gender (Hypothesis 2). Hypothesized relationships can be seen in our model in Figure 1.

Hypothesis 1: Individual’s gender will affect perception of and reaction to trolling.

Hypothesis 1a: Individual’s gender will affect perception of online trolling.

Hypothesis 1b: Individual’s gender will affect reaction to online trolling.

Hypothesis 2: Individual’s perception of and reaction to online trolling will vary by trolls’ gender.

Hypothesis 2a: Individual’s perception of online trolling will vary by trolls’ gender.

Hypothesis 2b: Individual’s reaction to online deviant behavior will vary by trolls’ gender.

Figure 1.

Reaction to and perception of troll’s behavior, motivation, and impact on online communities.

Various motives for online trolling have been identified (Downing, 2009; Shachaf & Hara, 2010), including activism, enjoyment, and malevolence. Activism or ideology ranges from “hacktivism” to the dissemination of information and the protest of violations of civil liberties (e.g., Jordan & Taylor, 1998; Kelly, Fisher, & Smith, 2006). Enjoyment motivates trolls, as some people find conflict and eliciting outrage or discomfort in others to be humorous (e.g., Danet, 2006; Downing, 2009; Turgeman-Goldschmidt, 2005). Overt malevolence or deviance also leads to trolling when actors seek to create conflict and cause harm (e.g., Shachaf & Hara, 2010). We hypothesized that individuals would perceive male and female motives differently (Hypothesis 3) and that an individual’s gender would affect perception of trolls’ motives (Hypothesis 4).

Hypothesis 3: An individual’s perception of trolls’ motives will vary by troll’s gender.

Hypothesis 4: An individual’s gender will affect perceptions of trolls’ motives.

There is also evidence that women perceive online deviance to be as negatively impactful as off-line deviance, whereas men perceived it to be less impactful (Whitty, 2005). We hypothesized that individuals would perceive the impact of trolling by male and female actors on the community differently (Hypothesis 5) and that an individual’s gender would affect perception of the impact of trolls’ behaviors (Hypothesis 6).

Hypothesis 5: An individual’s perception of the impact of trolling on the community will vary by trolls’ gender.

Hypothesis 6: An individual’s gender will affect perception of the impact of trolling on the community.

Context

Context, as the community in which interactions occur and the characteristics of that environment, is highly important in shaping and impacting sociotechnical interactions and must be taken into consideration when evaluating group outcomes and norms (Kling & Star, 1997). However, the relationship between the context and deviant behaviors has primarily been explored indirectly.

In research on hackers and trolls’ motivations (e.g., Jordan & Taylor, 1998; Shachaf & Hara, 2010; Turgeman-Goldschmidt, 2005), the context of specific behaviors seems to impact perceptions of motivations. For example, motivations have been identified as ideological when individuals introduce controversial topics into online news discussions (Kelly et al., 2006), yet are sometimes interpreted as evidence of loneliness when interjected into apolitical discussions, because people simply want to talk (Denegri-Knott & Taylor, 2005). Motivations can be characterized along internal and external lines, with internal motivations referring to psychological, cognitive, and emotional drivers, and external motivations referring to social and environmental drivers of behavior (e.g., Raban & Harper, 2008; Rafaeli & Ariel, 2008; Yee, 2006). Specific contexts have been correlated with specific motivations. For example, Wikipedia contributions are often motivated by internal factors (e.g., Rafaeli & Ariel, 2008); contributors share time and knowledge based on altruism and ideology (Nov, 2007). Social question and answer (Q&A) contributions are often externally motivated (e.g., Gazan, 2010; Raban & Harper, 2008); respondents seek to be seen as experts, negotiating status and control of particular topics (Harper, Raban, Rafaeli, & Konstan, 2008). Online game players are motivated internally and externally (e.g., Downing, 2009), seeking social negotiation (e.g., Downing, 2009; Yee, 2006) as well as challenge (Downing, 2009), achievement, and immersion (Yee, 2006). Haythornthwaite (2009) further suggests that motivations to contribute differ in essence by context, and she specifically differentiates between motivation to contribute online communities and crowds.

For normative and deviant motivations to contribute, different behaviors often result from the same motivation (Elliott & Dweck, 1988). It is possible that other motives exist for online deviance, which are not similar to the motives of normative behaviors, just as it is possible that behaviors, which are perceived as deviant, may in fact be the product of genuine confusion (Danet, 2006) or curiosity satisfied through experimentation (Kirman et al., 2012; Utz, 2005). As a result of the situational nature of perception and the connections between motivations and communities, we anticipate that individuals will perceive a troll’s motives differently based on the specific context in which the behavior is observed (Hypothesis 7).

Hypothesis 7: An individual’s perception of trolls’ motives will vary with context.

Method

This study employed context-specific scenarios to examine variations between different online communities’ perception of trolling motivation, exploring the role of context in motivation for online trolling. To examine how gender affects perception of and reaction to deviant behavior, the study manipulated the apparent gender of trolls in the scenarios. Though the nature of online communities allows for gender deception (Utz, 2005), participants perceive user names and avatars as representative of gender, which is sufficient for consideration of perception, rather than reality. Experimental methods and use of scenarios have rarely been applied in studies of online deviance and discrimination (Shachaf, Oltmann, & Horowitz, 2008; Utz, 2005; Whitty, 2005), despite their usefulness in isolating variables.

Data were collected using this unobtrusive method, so that participants were unconscious of the study and could not self-censor their responses (Bushman & Bonacci, 2004; Shachaf et al., 2008). Unobtrusive methods are particularly useful when sensitive variables, such as gender, are involved. So as not to make participants conscious of gender in providing different responses, the nature of the study was revealed only after responses were logged (Bushman & Bonacci, 2004). Participants were then debriefed with a short description of the manipulation and provided the opportunity to opt out of the study at this point.

Participants (n = 100) were college students at Midwestern universities, recruited through e-mail listservs, a digital bulletin board, and Twitter. Participants were presented with three brief scenarios, each describing an online interaction with a troll, and after completing an online survey through a Qualtrics interface (Qualtrics Labs, Inc., 2012), were asked to respond to a demographic survey that included gender, which was directly relevant to our study design.

Three scenarios were developed in order to limit the impact of other variables (Utz, 2005; Whitty, 2005). The value of using scenarios in this type of inquiry is the presentation of highly controlled data to drive individual responses, making participant responses more comparable (Utz, 2005; Whitty, 2005). Each of the scenarios described interaction in one of the three popular online communities: Wikipedia, an online encyclopedia; Yahoo! Answers, an online forum for question answering; and League of Legends, an online gaming site.

These specific cases represent good test sites because of their popularity, familiarity among the target population, established identities as communities, and documented experiences with trolling. Each community represents the most highly used platform within their domain; League of Legends is the most popular massively open online game, with 67 million players each month and an average of 27 million players per day (Tassi, 2014), while Wikipedia attracts hundreds of millions of visitors per month, as the web’s ninth most popular website (Javanmardi, Ganjisaffar, Lopes, & Baldi, 2009) and Yahoo! Answers continues to grow in use as social Q&A proliferates (Liu & Agichtein, 2008). Furthermore, each community has been studied for elements of community development and function (Javanmardi et al., 2009; Liu & Agichtein, 2008; Tassi, 2014). Trolling is also well documented within the Wikipedia community (e.g., Shachaf & Hara, 2010), in online forums (e.g., Herring, Job-Sluder, Scheckler, & Barab, 2002), and in online gaming communities (e.g., Downing, 2009). The three scenarios are:

Scenario 1: You contribute regularly to Wikipedia posts about X and participate in the discussions between contributors about what content is appropriate on pages in the area of X. A new contributor, Todd/Emily/AbcD, has changed information on many of these pages, without providing sources or explanations, causing a lot of debate between contributors about the new and altered content.

Scenario 2: You are a regular contributor to Yahoo! Answers, a Q&A site, focusing on feeds on subjects related to X, which you know a lot about. Lately, a new user, Todd/Emily/AbcD, has posted polarizing political comments as answers to questions related to X. As a result, combative political discussions, comprising dozens of posts, have developed, despite the questions and feeds topic as related to X.

Scenario 3: You regularly play League of Legends, a multiplayer online video game, and frequently interact with the same group of people. A new member, Todd/Emily/AbcD, has joined and has interrupted many members by asking questions during time-sensitive in-game activities.

Following each scenario were five Likert-type scale questions and a sixth question asking participants to identify all possible motivations. The first, “What do you think about Todd/Emily/AbcD’s behavior?” sought to assess what individuals thought the troll’s behavior was. The second and third “What would you do?” and “How do you think the community should respond?” evaluated reactions. The fourth assessed individuals’ perceptions of the impact of the behavior by asking, “What is the impact of this behavior on the community?” The fifth “How do you feel about this behavior?” asked for personal perceptions of the behavior. The sixth question addressed perceptions of the troll’s motivations: “Why do you think Todd/Emily/AbcD acted this way?”

For the sixth question, the multiple response question, we listed potentially relevant motivations based on the literature on trolling: loneliness as a desire to belong (Downing, 2009; Krappitz, 2012; Maratea & Kavanaugh, 2012), malevolence (e.g., Shachaf & Hara, 2010; Turgeman-Goldschmidt, 2005), humor (Krappitz, 2012), instigation (e.g., Hardaker, 2010), and ideology (e.g., Jordan & Taylor, 1998). We also allowed individuals to provide alternative motivations to those presented to them.

Each participant received a version of the survey that included three different scenarios (with three different user names). To avoid confounding variables, the counterbalanced method controlled for troll name’s impact. Each troll user name and each scenario were randomly paired.

Addressing the first six hypotheses, data by scenario were condensed to analyze variation in responses by gender of individuals and aggregate responses’ variation by gender of trolls. However, to test the seventh hypothesis, variation by scenario, we examined both statistically and qualitatively. A chi-square test of independence was performed for each hypothesized relationship. Cross tabulation of user names, scenarios, and genders was applied in order to appropriately test for statistically significant relationships. In addition, qualitative analysis of the open-ended question, asking participants to identify alternative motivations, used binary assessments of constructive versus destructive motivations. The responses to open questions, from 17 of the 100 participants, were used to provide nuance and illustrate the ways in which participants understood trolling motivations; these responses were not used in hypothesis testing.

Interpretation of the results was guided by synthesis of additional literature on gender norms, perception, reaction, and context, so as to identify theoretical constructs consistent with our findings to be woven into the existing online deviance framework. Discussion and interpretation of all results are firmly grounded in a broad literature.

One of the study’s limitations is that the scenarios do not reflect the complexity or dynamic context of reality. Yet they were carefully designed to assess particular variables and control for the impact of other variables; this trade-off allowed us, at the most basic level, to assess correlation in a controlled setting. Another limitation is that the participants were mostly female graduate college students at a single Midwestern university, reacting to hypothetical scenarios and not real participants in the specific online communities. Yet the demographics of participants, as socially privileged, young women, do provide an advantage in matching those of social online community members who tend to be more affluent, educated, young, and female, as they have greater access to technology (Duggan & Brenner, 2013).

Findings

Participants (n = 100) represented a total of 35 different degree majors; 62% were graduate students, 27% were undergraduates, and 11% represented other academic statuses, including certificates, continuing students, and postdoctoral scholars. Seventy-five percent of participants were female, 23% were male, and 2% responded as other. The vast majority (94%) attended the same Midwestern university; their age ranged from 18 to 53, with an average of 26.

We hypothesized that perceptions of (Hypothesis 1a) and reaction to (Hypothesis 1b) trolling would vary by gender of the individual participant (Hypothesis 1). The chi-square test was significant for reaction to trolling, Pearson χ²(4, N = 600) = 8.664, p = .001, but not significant for perception of trolling Pearson χ²(4, N = 400) = .967, p = .915; thus, Hypothesis 1 was partially supported. Men tended to engage actively in discussions with trolls or block them while women’s reactions varied, and more of them preferred to ignore the discussion or discourage debate instead of blocking or confronting the troll (Table 1). Men and women perceived trolling similarly, with most emphasis on the negative end of the Likert spectrum. As can be seen in Table 1, men and women did not perceive trolling differently but reacted differently to trolling behaviors.

Table 1.

Perception of and Reaction to Trolling by Gender (Hypotheses 1 and 2).

Perception by Gender of Individual	Positive	Somewhat Positive	Neutral	Somewhat Negative	Negative
Hypothesis 1a (Qs 1, 4, and 5)	Positive	Somewhat Positive	Neutral	Somewhat Negative	Negative
Female	3.3%	10.7%	17.7%	40.7%	27.6%
Male	4.3%	11.6%	19%	38%	27%
Reaction by gender of individual	Active participation	Passive participation	Ignoring	Passive opposition	Active opposition
Hypothesis 1b (Qs 2 and 3)	Active participation	Passive participation	Ignoring	Passive opposition	Active opposition
Female	24%	26.5%	20%	17.5%	11.5%
Male	33.5%	25.5%	12.5%	8.5%	20.5%
Perception by troll’s gender	Positive	Somewhat positive	Neutral	Somewhat negative	Negative
Hypothesis 2a (Qs 1, 4, and 5)	Positive	Somewhat positive	Neutral	Somewhat negative	Negative
Emily	5.6%	14.6%	18.3%	36.3%	25%
Todd	2%	8.6%	14.6%	45.6%	29%
AbcD	3%	8.6%	21%	38.6%	28.6%
Reaction by troll’s gender Hypothesis 2b (Qs 2 and 3)	Active participation	Passive participation	Ignoring	Passive opposition	Active opposition
Emily	30%	25.5%	14%	14.5%	15%
Todd	25.5%	27.5%	18.5%	27%	11.5%
AbcD	23%	24.5%	22.5%	15.5%	14.5%

We further hypothesized that individual participants would perceive the same trolling behavior exhibited by men and women differently (Hypothesis 2a) and therefore reaction to online trolling would vary by trolls’ gender (Hypothesis 2b). The chi-square test was significant for perception of trolling Pearson χ²(8, N = 900) = 20.775, p = .008, but not significant for reaction to trolling Pearson χ²(8, N = 600) = 7.679, p = .465.

In addition, we hypothesized that individual participants would perceive men and women trolls’ motives differently (Hypothesis 3) and that individuals’ gender would affect perception of motivation (Hypothesis 4). The relation between trolls’ gender and perception of motivation was significant Pearson χ²(14, N = 790) = 23.774, p = .049. Hypothesis 3 was thus supported. Participants perceived Todd and AbcD to be motivated by malevolence, humor, and instigation at a higher frequency than Emily who was perceived to be comparatively more ideological (Table 2). However, some motivations, such as loneliness, confusion, and curiosity, were assigned equally to trolls regardless of their gender. The relation between perceived motivation for trolling and an individual’s gender was not significant, Pearson χ²(7, N = 516) = 3.411, p = .845. Hypothesis 4 was thus not supported. Yet female participants were more likely to perceive confusion, curiosity, and instigation as motivations for trolling; male participants were more likely to perceive malevolence as motivation for trolling (Table 2).

Table 2.

Perceived Motivation to Troll by Gender.

Hypothesis 3 (Q6)	Loneliness (%)	Malevolence (%)	Confusion (%)	Humor (%)	Curiosity (%)	Instigation (%)	Ideology (%)	Other (%)
Emily	39	16	44	10	38	34	52	10
Todd	34	36	45	26	34	46	41	14
AbCD	36	37	44	26	36	45	41	6
Hypothesis 4 (Q6)
Female	36	28	46	20	38	42	45	10
Male	34	34	37	23	27	37	42	8

Note. Hypothesis 3 = troll’s gender; Hypothesis 4 = individual gender.

Hypotheses 5 and 6, respectively, anticipating the influence of trolls’ and individuals’ genders on the perceived impact of trolling on a community, were tested based on answers to Q4. The relation between perception of the impact of online deviant behavior and trolls’ gender was not significant, Pearson χ²(8, N = 300) 8.625, p = .375. Hypothesis 5 was thus not supported. Likewise, the relation between perception of the impact of online trolling and individuals’ gender was not significant, Pearson χ²(4, N = 200) = 1.299, p = .862; Hypothesis 6 was not supported. While male’s trolling was perceived to have a slightly more destructive impact on the community compared with that of female’s or those with gender-neutral names, the impact of trolling on the community did not differ significantly by trolls’ gender nor by participants’ gender (Table 3).

Table 3.

Perceived Impact of Trolling by Gender.

Hypothesis 5 (Q4)	Constructive (%)	Slightly Constructive (%)	No Effect (%)	Slightly Destructive (%)	Destructive (%)
Emily	12	19	11	44	14
Todd	4	16	11	53	16
AbcD	5	16	16	45	18
Hypothesis 6 (Q4)
Female	7	17	12	49	15
Male	9	17	13	42	19

Note. Hypothesis 5 = troll’s gender; H6 = individual gender.

Finally, we hypothesized that the context of different scenarios may lead to different perceptions of motivations (Hypothesis 7). Table 4 presents the percentages of participants who identified each motivation by scenario. There was significant variation by scenario, Pearson χ²(14, N = 790) = 254.949, p = .000. For example, the political behaviors on Yahoo! Answers were not perceived as motivated by confusion, while both other scenarios were. Responses also illustrate the higher perception of instigation and ideology in motivating political online communication than in new members of online gaming communities asking questions. There is also evidence that participants interpret asking question in an online game as more motivated by confusion and curiosity than in the other contexts.

Table 4.

Perceived Motivation by Context.

Scenario	Wikipedia (%)	Yahoo! Answers (%)	League of Legends (%)
Loneliness	33	21	55
Malevolence	33	48	8
Confusion	49	7	77
Humor	29	26	7
Curiosity	20	21	67
Instigation	41	78	6
Ideology	56	62	16
Other	16	8	6

The open-ended responses, provided by a small subset of participants (17 of the 100), included a variety of alternate explanations. Some participants asserted that motivations could not be accurately inferred, responding, for example, “No way to know.” One participant felt that all the motivations were reasonable in all three scenarios. Others identified additional motivations for the scenario of deviant behavior on Wikipedia, including “conceit” or “arrogance,” thoughtlessness, wanting to contribute expertise without membership, and lack of understanding of citation rules or “the editorial culture.” For Yahoo! Answers, two users explicitly identified the deviant as a troll, of which one argued that “The only way to combat a troll is to ignore”; another stated that the trolling had occurred because the particular user was “a jerk with no consequences” and yet another stated that “It would be helpful to see a sample of his posts, to determine if they are inflammatory/instigating in nature.” For the League of Legends scenario, participants emphasized a lack of understanding for new users of complex games as well as offering “ignorance” and psychological problems as explanations.

Discussion

Our model illustrates the impact of gender and context on online trolling, with an emphasis on representation of the statistically significant relationships between variables (Figure 1). As can be seen in our model, and in line with our expectations, context impacts perception of motivation to troll. The findings also show that men and women reacted differently to trolling, but both men and women perceived men trolls differently from women trolls; in other words, reactions varied by participants’ genders, whereas perceptions varied by the gender of the troll. This echoes results of studies in off-line contexts, which show that women deviants are viewed as less socially problematic or dangerous than men deviants (Schnittker, 2000); in this sense, comparison of our findings to existing literature, which grounds this discussion, helps support our claims of implications identified.

Results appear largely consistent with contemporary discussions in American society regarding gender, in that women are more socially passive than men in situations of conflict (Herring, 2008), whereas implicit biases are similarly perceived by all genders (Erickson, Crisnoe, & Dornbusch, 2000). This suggests that implicit gender expectations impact behavior and perception, but we need to better understand how other contextual aspects, such as gender-symmetric relationships (Schnittker, 2000) and group identification (Erickson et al., 2000; Özden, 2008), may contribute to inconsistencies in this correlation.

Prior to this study, little research had addressed the impact of deviant, antisocial, and nonnormative behaviors on online communities. While theorization about social and institutional norms pervaded other areas of study on the social aspects of information technology in very useful ways (e.g., Kling & Iacono, 1984), theories concerning trolling only conceptualized types of deviant behaviors and interventions (Suler & Phillips, 1998), without addressing underlying social psychological aspects of deviants and fellow community members. The results of this study serve to inform our understanding of online trolling. In the following sections, we discuss the impact of gender (Gender section) and context (Context section) on trolling as well as how the integration of these variables enhances Suler and Phillips’ typology. These sections will be organized by research question; in each section, we discuss our findings in light of existing scholarship and highlight the theoretical implications for Suler and Phillips’ typology.

Gender

Does gender affect perception of men’s and women’s trolling, motivations, and impact on the community?

The differences between men’s and women’s perceptions of trolling as identified in our study enhance our understanding of the relationship between gender and online deviant behavior in general and online trolling in particular. Previous studies in off-line contexts have found girls to be less likely to exhibit deviant behaviors than boys and to have stronger group affiliation (Özden, 2008). Yet, unlike off-line expectations that women contribute positively to the group (Özden, 2008), in our study, the impact of women trolls online was not perceived more positively. Future studies should empirically examine it further, while comparing behavioral expectations in online and off-line contexts and controlling for other variables.

Gender differences exist in normative online contributions; studies suggest that women are more active in blogging and Q&A communities and that women are “more interested in the social aspects” and seek to build communities constructively (Lu et al., 2010, p. 403). Furthermore, from a gender-normative perspective, certain attitudinal attributes are associated with gender. Specifically, perceptions of gendered behavior lead individuals to subconsciously anticipate future gendered dispositions (Deaux & Major, 1987); previous experimental research illustrates that individuals explain the same behaviors differently for men and women based on these implicit biases (Chartrand & Bargh, 1999). Our finding that women trolls were perceived to have less negative motivations, with malevolence and instigation less often identified than for men or gender-neutral trolls is in line with prior research.

These behavioral expectations found in our study also shed light on social norm enforcement within online communities. Norms are actively enforced through men’s assertive behavior and women’s nonconfrontational behavior (Lester, 2011), implicitly shaping gender biases and expectations for acceptable and deviant behaviors (Erickson et al., 2000). Norms are passively enforced through unspoken standards of behavior, as gendered expectations of others, met or unmet, shape perception and serve to clearly identify the expected behavior of new members in active, socially rich communities (Widen-Wulff et al., 2008).

Based on the results that perception of behaviors of and motivation for trolling varies by the gender of the troll, conceptualization of deviance can be enhanced. Gendered perceptions of identical trolling behaviors map compatibly onto the spectrum for deviance in Suler and Phillips’ (1998) typology, with female perpetrators characterized at the mild end and male perpetrators as responsible for more severe deviance. In this sense, this study builds a gendered layer onto the existing theory, which is useful, as Suler and Phillips acknowledged the limited number of and access to female deviants online.

However, our findings also imply disconnect between the impact of deviant actors as characterized by the typology and perceptions of their impact based on the troll’s gender. For example, Suler and Phillips (1998) define a “revolutionary” to be a user who stirs up controversy by espousing extreme political views, and we found that a troll assuming the goals and behaviors of a revolutionary, as in the scenario of political argumentation within online Q&A, is viewed as less severe if expressed by female than by male. Specifically, trolls’ gender affected perceived motivations in these cases. Individuals who responded to our survey more often identified Emily’s deviant acts as those of “clueless newbies,” prescribing motivations of confusion and curiosity to female trolls, even though the behavior described was both the same as for a male troll and prototypical of a revolutionary. The responses to our survey are, with respect to female trolls, consistent with Suler and Phillips’ (1998) typology of mildly deviant actions, thereby illustrating the gendered dimension of deviant behavior.

Does gender affect reaction to men’s and women’s trolling?

Research on reactions to deviance, online and off-line, provides evidence that individuals react differently to men and women (Kunkel & Joyce Mccari, 1998). For example, Schnittker (2000) found that men and women were both more likely to interact with women deviants than men deviants, because they were perceived as less dangerous. Surprisingly, our results did not show differences in reaction based on troll’s gender, but rather only by participant’s gender; nor did it show differences in the perception of impacts on the community based on troll’s gender. In other words, reaction to trolling by men and women did not differ significantly. Previous research reveals that some commonly used conflict management behaviors in face-to-face interaction did not produce similar results in virtual groups (Montoya-Weiss, Massey, & Song, 2001). Thus, a possible explanation is disinhibition in computer-mediated groups and online mediation; online disinhibition has been shown to reduce aversion to conflict (Suler, 2004), which could make people equally willing to interact with men and women deviants due to social distance. Yet, further research should test this possible explanation by manipulating medium of interaction while testing the same deviant behavior.

Results from our study indicate that perceptions of trolling behaviors vary by gender, in terms of both reacting to and interaction with other members of the community, and that norms and expectations are actively produced by gendered behaviors when men and women demonstrate different reactions to the same trolling behavior. In a given context, appropriate social behaviors serve as precedent by which future actions are judged (Widen-Wulff et al., 2008). Previous research has evaluated the online construction of gendered and sexual norms (e.g., Siibak & Hernwall, 2011), as well as the construction of online norms (e.g., McLaughlin & Vitak, 2012), but has not substantively addressed the role of gender in construction or enforcement of online norms. Gendered norms are significant to the analysis of the role of gender in determining appropriate reactions to norm violation, when individuals perceive norms to be broken, and how people assess the impact of broken norms.

In the context of deviant behaviors, Suler and Phillips’ (1998) conventions for intervention can be further enhanced through the integration of gendered reaction to online trolling. Based on the result that men are more likely to confront deviant users, social construction of masculinity appears consistent with many of the attributes Suler and Phillips describe as fundamental to successful interpersonal intervention. They suggest a forceful and assertive presence in recognizing deviance and bringing community member awareness to the attention of the troll without engaging in public reprimand that may lead to further undesirable behaviors. The direct interaction between individuals, rather than public shaming or community outrage, favored by male participants in our study is perhaps the best mechanism to stop trolls. This consistency between the typology of responses (Suler & Phillips, 1998) and male participants professed intended reactions is logical in the male-dominated spaces on the Internet.

In contrast, Suler and Phillips (1998) also emphasize that being calm and making friendly attempts to socialize and rehabilitate deviant users in a manner likened to “social workers’” (Suler & Phillips, 1998, p. 290), a stereotypically female profession. In terms of individual and community responses, the impulses identified by respondents in our study, better aligned men with Suler and Philips’ typology, the few women who would have interpersonally responded may have had better success in an actual attempt to stop or rehabilitate a troll. This raises questions as to whether men or women have better success in mediating trolling behaviors. While our study did not focus on specific interventions or reaction to trolling, future research should assess reaction effectiveness.

Context

How does context impact perception of and motivations for trolling?

Communities have different expectations regarding membership, contribution, and standards of behavior. Our data revealed that context, defined by the specific community, led to different perceptions of motivation for online trolling. Similar norm infringements were interpreted differently based on the regular practices and nature of communities. For example, the trolling behavior in the context of Yahoo! Answers was viewed as more inflammatory than in the contexts presented within other contexts. Trolling behavior in this context was more likely to be interpreted as motivated by malevolence and instigation, because the nature of the trolling behavior, which included political comments, was too far removed from community expectations. While participants identified socially negative motivations and “ignorance” for noncompliance in the Q&A community (Yahoo! Answers) and online gaming environments, some reactions to the same behaviors in the Wikipedia context were different; expertise on Wikipedia was, comparatively, as important as social aspects with regard to contributions. In this sense, because expertise was valued, some participants viewed the trolling behavior as positively motivated, positive contribution; they stated that expertise and knowledge were valuable contributions without expecting integration in the community. Furthermore, specific aspects of trolling behaviors lead to differences in perception, as indicated by the participant who said that examples of additional posts would be necessary to ascertain motivations for the behavior of an individual.

These findings are interesting, in light of previous research, because of (1) the contextual variation in perception is consistent for both normative and deviant motivations; (2) the consistency between trolls’ professed motivations and perceptions of the community members; and more generally (3) the evidence of impact of context on perceptions of trolling.

1) Contextual variation in perception is conserved for both normative and deviant motivations. Previous scholarship has identified motivation for normative contributors to various online communities, and our research extends this by showing that community members identify similar motivations for antinormative contributions. Specifically, previous research has indicated that motivations to contribute to Wikipedia include internal self-concepts (Nov, 2007; Rafaeli & Ariel, 2008), whereas contributions to Yahoo! Answers are motivated by external, social–capital and influence-based factors (Gazan, 2011; Raban & Harper, 2008). Our results show that trolls in these communities are perceived to be motivated both by internal (ideology, confusion, and curiosity) and by external factors (loneliness, malevolence, and instigation).

Ideology and curiosity (internal motivations) as well as loneliness, instigation, and malevolence (external motivations), were most frequently identified in all three scenarios. This is interesting because curiosity and loneliness lead to normative behaviors as well (Gazan, 2011; Nov, 2007). What is evident from the results, more generally, is that (a) internal motivations are identified for deviant Wikipedia contributions, consistent with previous research on all contributions (Nov, 2007; Rafaeli & Ariel, 2008); (b) external motivations are identified for deviant Q&A participation, just as has been identified for normative participation (Gazan, 2011; Raban & Harper, 2008); and (c) a combination is identified for online gaming communities, as has been identified for gamers overall (Downing, 2009). In our survey, we included deviant motivations, such as malevolence, which have not been identified as motivators for normative contributions but rather was drawn from conceptualizations of deviant behavior (e.g., Suler & Phillips, 1998).

2) The identified motivations also imply that community members perceive motivations consistently with deviants’ professed motivations. For example, Downing (2009) identified motivations for online game deviance as loneliness and curiosity, which was echoed in the online game-playing scenario in our study; however, the other major motivation Downing identified was malevolence, which in our study was rarely identified. This similarity between intention and perception was certainly not clear prior to this study, and it is difficult to explain, because decades of social psychology research indicate that people are relatively inaccurate when guessing why individuals do specific things (Ross, Greene, & House, 1977). Future analysis may examine the possibility that contextual variables in face-to-face versus online interaction affect the accuracy of perception of motives differently.

3) Our study provides empirical supports for the impact of context on perceptions that could be only inferred from prior research. For example, Gazan (2010, 2011) has identified socially affective variables, such as control or belonging, as motivations for Q&A participation, which are highly correlated with loneliness and social impact; some of our participants responded that this could also drive deviant participation, and many identified instigation, which is a form of social impact and control, as a motivation. Perceptions of antisocial Wikipedia behaviors are also consistent with prior research on Wikipedia trolls; Shachaf and Hara (2010) identified malevolence, humor, and instigation as motivations, which our study participants identified as well. Results within these contexts were also highly specific to the behaviors, with emphasis on ideology, which reflects the nature of the comments and interactions.

Contextual variation in motivations was anticipated based on previous empirical studies (Williams, 2000). Many of the differences in motivations between scenarios can be explained by specific contextual attributes. For example, when users promote political arguments on Yahoo! Answers forums without political implications, individuals perceive this as more ideological than asking disruptive questions during a game, because politics are ideological, while lack of understanding and even provocation, as a child would repetitively ask “why,” are rarely ideological rhetorical devices. Less intuitive is the variation in the perception of confusion, which was very infrequently perceived in the case of off topic remarks on Yahoo! Answers when compared with the other scenarios.

Furthermore, our research supports claim of social informatics that technologically mediated interactions are socially shaped (e.g., Kling & Star, 1997). Not only does the context of behavior matter for external motivations (Gazan, 2011; Harper et al., 2008) and features of online environments (Suler, 2004) but also for internal motivations (Downing, 2009; Nov, 2007). This is particularly meaningful when considering that within a context, different behaviors may manifest from the same factors (Elliott & Dweck, 1988) due to differences in internal variables rather than the external social environment. Previous research has suggested relationships between context and perception of motivations (e.g., Whitty, 2005) and context and perception of behaviors (e.g., Chartrand & Bargh, 1999), but future research should further explore variation by context in perception of impact specifically for trolling.

Finally, what the variation in perceived motivations illustrates is that behaviors are understood differently in different contexts, making Suler and Phillips’ (1998) characterizations of deviant acts contextually dependent. While Suler and Phillips acknowledge that norms and standards about what is acceptable and unacceptable behavior vary from one culture to another, they did not describe specifically which behaviors a particular community would consider to be deviant. While their typology places question asking and confusion in a mild category of nuisance attributable to newcomer status, some communities (e.g., online gaming communities) viewed more negatively (Downing, 2009). From our results, it is also evident that the incendiary and complex nature of “revolutionaries,” as defined by Suler and Phillips to encompass political instigation inappropriate to the community, was viewed by less than half of participants as descriptive of the motivations for a political troll on Wikipedia.

Conclusion

This study begins to elucidate the role of gender and context in understanding trolling in online communities. Specifically, evidence is identified that an individuals’ gender is related to their reaction to behavior, while a troll’s gender relates to public perception of trolling and motivations. The significance of these findings is their potential to ground theorization on trolling in terms of prevention, perception, and reaction. Gendered behaviors produce expectations, which are formalized as norms (e.g., Erickson et al., 2000); women are perceived as more constructive than men in their behavioral variation from norms (Özden, 2008), and men are more willing to confront violators (Lester, 2011).

Our research validates and enhances aspects of Suler and Phillips’ (1998) framework of online deviant behavior, through its demonstration of a spectrum of behaviors and characteristics of interventionist reactions to deviance. Perhaps more importantly, our study projects gendered and contextual aspects onto this spectrum, thereby enhancing this typology of deviant behaviors. Specifically, mild forms of deviance are consistent with female behaviors; male reactions to deviance are akin to the model for interpersonal intervention; and the same behavior can be conceptualized differently in different contexts. The integration of these variables is not trivial, given their significance to deviant behaviors and social interaction with trolls in online communities. Gender and context are also significant in that the original typology at once considered them, but not as variables and therefore did not integrate them into the characterizations they presented.

This research supported our expectation that context of online trolling impacts perception of trolls’ motivations. Specifically, as individual communities establish patterns of normative behaviors, the context of community standards makes certain aspects of behaviors socially acceptable or socially deviant (Chartrand & Bargh, 1999). For example, our results indicate that aggressive behaviors are relatively unwelcome in online Q&A communities and are seen as ideologically motivated, rather than something humorous that is acceptably done in good fun, as reported for gaming communities (Downing, 2009; Yee, 2006).

Footnotes

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

References

Adam

A. E.

(2009). The gender agenda in computer ethics. In Himma

Tavani

H. T.

(Eds.), Handbook of information & computer ethics (pp. 589–619). Hoboken, NJ: John Wiley.

Baker

(2001). Moral panic and alternative identity construction in Usenet. Journal of Computer-Mediated Communication, 7. Retrieved from http://onlinelibrary.wiley.com/doi/10.1111/j.1083-6101.2001.tb00136.x/full

Barak

(2005). Sexual harassment on the Internet. Social Science Computer Review, 23, 77–92.

Bushman

B. J.

Bonacci

A. M.

(2004). You’ve got mail: Using e-mail to examine the effect of prejudiced attitudes on discrimination against Arabs. Journal of Experimental Social Psychology, 40, 753–759.

Chartrand

T. L.

Bargh

J. A.

(1999). The chameleon effect: The perception–behavior link and social interaction. Journal of Personality and Social Psychology, 76, 893–910.

Danet

(2006). “What a wonderful do-krav this!” Language, culture, and gender in an Israeli flame event. Paper presented at the Fifth Annual Conference, Israel Association for Language and Society, Open University, Ra’anana, Israel.

Deaux

Major

(1987). Putting gender into context: An interactive model of gender-related behavior. Psychological Review, 94, 369–389.

Denegri-Knott

Taylor

(2005). The labeling game: A conceptual exploration of deviance on the internet. Social Science Computer Review, 23, 93–107.

Downing

(2009). Attitudinal and behavioral pathways of deviance in online gaming. Deviant Behavior, 30, 293–320.

10.

Duggan

Brenner

. (2013). The demographics of social media users—2012. Pew Research Center. Retrieved from http://pewinternet.org/Reports/2013/Social-media-users.aspx

11.

Elliott

E. S.

Dweck

C. S.

(1988). Goals: An approach to motivation and achievement. Journal of Personality and Social Psychology, 54, 5–12.

12.

Erickson

Crosnoe

Dornbusch

(2000). A social process model of adolescent deviance: Combining social control and differential association perspectives. Journal of Youth and Adolescence, 29, 395–425.

13.

Gazan

(2010). Microcollaborations in a social Q&A community. Information Processing and Management, 46, 693–702.

14.

Gazan

(2011). Social Q&A. Journal of the American Society for Information Science & Technology, 62, 2301–2312.

15.

Hardaker

(2010). Trolling in asynchronous computer-mediated communication: From user discussions to academic definitions. Journal of Politeness Research-Language Behaviour Culture, 6, 215–242.

16.

Harper

Raban

Rafaeli

Konstan

J. A.

(2008). Predictors of answer quality in online Q&A sites. In Proceedings of the Annual International ACM SIGCHI Conference on Human Factors in Computing Systems Proceedings (Florence, Italy, April 5–10, 2008). SIGCHI ’08 (pp. 865–874). New York, NY: ACM.

17.

Haythornthwaite

(2009, January). Crowds and communities: Light and heavyweight models of peer production. Proceedings of the 42nd Hawaii International Conference on System Sciences. Los Alamitos, CA: IEEE Computer Society. Retrieved from http://hdl.handle.net/2142/9457

18.

Herring

(2008). Gender and power in on-line communication. In Holmes

Meyerhoff

(Eds.), Handbook of language & gender (p. 202). Hoboken, NJ: Wiley-Blackwell.

19.

Herring

Job-Sluder

Scheckler

Barab

(2002). Searching for safety online: Managing “trolling” in a feminist forum. Information Society, 18, 371–384.

20.

Javanmardi

Ganjisaffar

Lopes

Baldi

(2009). User contribution and trust in Wikipedia. In Proceedings of The 5th Annual IEEE International Conference on Collaborative Computing: Networking, Applications and Worksharing, 1–6, November 11–14. Washington, DC: IEEE.

21.

Jordan

Taylor

(1998). A sociology of hackers. Sociological Review, 46, 757–780.

22.

Kelly

J. W.

Fisher

Smith

(2006). Friends, foes, and fringe: Norms and structure in political discussion networks. In Proceedings of the 2006 International Conference on Digital Government Research (San Diego, CA, May 21–24) (pp. 412–417). San Diego, CA: Digital Government Society of North America.

23.

Kirman

Lineham

Lawson

(2012). Exploring mischief and mayhem in social computing or: How we learned to stop worrying and love the trolls. In Proceedings of the 2012 ACM Annual Conference extended abstracts on Human Factors in Computing Systems Extended Abstracts (pp. 121–130) (CHI EA ’12). New York: NY: ACM.

24.

Kling

Iacono

(1984). Computing as an occasion for social control. Journal of Social Issues, 40, 77–96.

25.

Kling

Iacono

(1989). The institutional character of computerized information systems. Information Technology & People, 5, 7.

26.

Kling

Star

(1997). Human centered systems in the perspective of organizational and social informatics (Chap. 5). In Huang

Flanigan

(Eds.), Human centered systems. National Science Foundation.

27.

Krappitz

(2012). Troll culture (PhD thesis). Stuttgart, Germany: Merz Academy College of Design, Art and Media.

28.

Kunkel

Joyce Mccari

(1998). Gender, residual deviance, and social control. Deviant Behavior, 19, 339–360.

29.

Lester

(2011). Regulating gender performances: Power and gender norms in faculty work. NASPA Journal about Women in Higher Education, 4, 142–169.

30.

Liu

Agichtein

(2008). On the evolution of the Yahoo! Answers QA community. In Proceedings of the 31st Annual International ACM SIGIR Conference on Research and Development in information Retrieval (Singapore, Singapore, July 20–24) (pp. 737–738). New York, NY: ACM.

31.

H.-P.

Lin

J. C.-C.

Hsiao

K.-L.

Cheng

L.-T.

(2010). Information sharing behavior on blogs in Taiwan: Effects of interactivities and gender differences. Journal of Information Science, 36, 401–416.

32.

Mahoney

J. L.

Stattin

(2000). Leisure activities and adolescent antisocial behavior: The role of structure and social context. Journal of Adolescence, 23, 113–127.

33.

Maratea

R. J.

Kavanaugh

P. R.

(2012). Deviant identity in online contexts: New directives in the study of a classic concept. Sociology Compass, 6, 102.

34.

McLaughlin

Vitak

(2012). Norm evolution and violation on Facebook. New Media & Society, 14, 299–315.

35.

Montoya-Weiss

M. M.

Massey

A. P.

Song

M. M.

(2001). Getting it together: Temporal coordination and conflict management in global virtual teams. Academy of Management Journal, 44, 1251–1262.

36.

Muhammad

(2000, December). Hacktivism. Ms magazine, 11, 74–76.

37.

Nov

(2007). What motivates Wikipedians? Communications of the ACM, 50, 60–64.

38.

Özden

Ö.

(2008). Self-control, gender, and deviance among Turkish university students. Journal of Criminal Justice, 36, 72–80.

39.

Puente

S. N.

(2011). Feminist cyberactivism: Violence against women, internet politics, and Spanish feminist praxis online. Continuum: Journal of Media & Cultural Studies, 25, 333–346.

40.

Puente

S. N.

Jiménez

A. G.

(2011). Inhabiting or occupying the web? Virtual communities and feminist cyberactivism in online Spanish feminist theory and praxis. Feminist Review, 39–54.

41.

Qualtrics Labs, Inc. (2012). Qualtrics (Version 12,018) [Computer software]. Provo, Utah: Qualtrics Labs.

42.

Raban

Harper

(2008). Motivations for answering questions online. In Caspi

Azran

(Eds.), New Media and Innovative Technologies (pp. 273–295). Beer-Sheva, Israel: Ben Gurion University of the Negev Press.

43.

Rafaeli

Ariel

(2008). Online motivational factors: Incentives for participation and contribution in Wikipedia. In Barak

(Ed.), Psychological aspects of cyberspace: Theory, research, applications (pp. 243–267). Cambridge, England: Cambridge University Press.

44.

Ross

Greene

House

(1977). The “false consensus effect”: An egocentric bias in social perception and attribution processes. Journal of Experimental Social Psychology, 13, 279–301.

45.

Sanfilippo

M. R.

Fichman

(2014). Social informatics: Milestones, opportunities, and challenges. In Fichman

Rosenbaum

(Eds.), Social informatics: Past, present, and future (pp. 31–55). Newcastle upon Tyne, England: Cambridge Scholars Publishing.

46.

Schnittker

(2000). Gender and reactions to psychological problems: An examination of social tolerance and perceived dangerousness. Journal of Health And Social Behavior, 41, 224–240.

47.

Shachaf

Hara

(2010). Beyond vandalism: Wikipedia trolls. Journal of Information Science, 36, 357–370.

48.

Shachaf

Oltmann

Horowitz

(2008). Service equality in virtual reference. Journal of the American Society for Information Science & Technology, 59, 535–550.

49.

Siibak

Hernwall

(2011). ‘Looking like my favourite Barbie’—Online gender construction of tween girls in Estonia and in Sweden. Studies of Transition States & Societies, 3, 57.

50.

Suler

(2004). The online disinhibition effect. Cyberpsychology & Behavior, 7, 321–326.

51.

Suler

Phillips

W. L.

(1998). The bad boys of cyberspace: Deviant behavior in a multimedia chat community. CyberPsychology & Behavior, 1, 275–294.

52.

Tassi

. (2014). Riot’s ‘League of Legends’ reveals astonishing 27 million daily players, 67 million monthly. Forbes. Retrieved January 27, 2014, from http://www.forbes.com/sites/insertcoin/2014/01/27/riots-league-of-legends-reveals-astonishing-27-million-daily-players-67-million-monthly/

53.

Turgeman-Goldschmidt

(2005). Hackers’ accounts: Hacking as a social entertainment. Social Science Computer Review, 23, 8–23.

54.

Utz

(2005). Types of deception and underlying motivation: What people think. Social Science Computer Review, 23, 49–56.

55.

Whitty

M. T.

(2005). The realness of cybercheating: Men’s and women’s representations of unfaithful internet relationships. Social Science Computer Review, 23, 57–67.

56.

Widen-Wulff

Ginman

Perttila

Sodergard

Totterman

A.-K.

(2008). Information behavior meets social capital: A conceptual model. Journal of Information Science, 34, 346–355.

57.

Williams

(2000). Virtually criminal: Discourse, deviance and anxiety within virtual communities. International Review of Law, Computers & Technology, 14, 95–104.

58.

Yee

(2006). Motivations for play in online games. CyberPsychology & Behavior, 9, 772–775.