How Do Silent Trolls Become Overt Trolls? Fear of Punishment and Online Disinhibition Moderate the Trolling Path

Abstract

Digital media allow users the ability to engage in and be exposed to trolling. Although many people may enjoy the occasional opportunity to witness others being trolled, a relative minority directly troll others, those whom we can label overt trolls. Nevertheless, features afforded on social media and online communities (e.g., likes, upvotes) make it accessible for people to positively react to and support trolling, becoming supportive trolls, a potential steppingstone into overt trolling. In the theoretical contexts of social cognitive theory and the bystander effect, we advance a model in which enjoyment of observing trolling prompts supportive trolling, which could then lead to overt trolling. Analyses of data from an online survey conducted in the United States (N = 604) show the positive link between enjoyment of observing trolling and supportive trolling is stronger among individuals with higher fear of punishment, while the subsequent link between supportive and overt trolling is stronger among those with higher online disinhibition. Our findings hold implications in understanding the effects of trolling on social media audiences and how trolling can be performed in nuanced ways.

Keywords

trolls disinhibition social cognitive theory bystanders

Using Goffman’s (1959) allegory of dramaturgical performance, social media have been compared to performance stages and exhibitions where curated artifacts are displayed for imagined audiences (Hogan, 2010). Web 2.0 platforms, including social media and online communities, allow users the opportunity to contribute to or evaluate content presented by others. On these platforms, users can also engage in antisocial behaviors, such as trolling. Pew Research Center findings suggest that 41% of U.S. adults acknowledge personally experiencing online harassment (Vogels, 2021). Furthermore, other research indicates that 64% of U.S. adults ages 25–34 years admitted trolling someone at some point (Gallegos, 2021). Among Australian teenagers, 24% reported being trolled while 13% admitted trolling others (Marrington et al., 2023). Because trolling is frequently performed in online environments, many internet users are exposed to these behaviors regularly. Some may even purposefully seek such trolling content for enjoyment. We argue that anyone can be a troll by enjoying the antisocial behaviors of others online.

Incorporating a theoretical framework derived from social cognitive theory (Bandura, 2001) and the interpersonal literature on bystanders, our study proposes a typology of online trolling and investigates how trolling bystander behaviors could turn into overt trolling (e.g., intentionally upsetting people online, sending others to shock websites; see Buckels et al., 2014; Craker & March, 2016). We are particularly interested in what Brubaker et al. (2021) described as silent trolls, or those who are “more than bystanders” and “enjoy watching the incivility play out in front of them” (p. 10). We propose that there are two types of silent trolls: inactive trolls who lurk and enjoy observing trolling and supportive trolls who utilize social media features (e.g., like, upvote) to support others’ trolling behaviors (see Table 1). We argue that supportive trolling could be an important steppingstone into overt trolling. In the context of social cognitive theory, people can become familiar with trolling behaviors they often observe and enjoy and develop behavioral intentions to duplicate such behaviors depending on the perceived reward or punishment (Bandura, 2001). In addition, the bystander effect (Latanè & Darley, 1968) could explain how people perceive their role as an audience to trolling, and whether they choose to support or discourage those behaviors.

Table 1.

Proposed Types of Trolling Behaviors.

	Silent trolling
	Inactive trolling	Supportive trolling	Overt trolling
Attitudes	Enjoyment of observing trolling	Enjoyment of observing trolling	Enjoyment of trolling
Behavior	Observe trolling	Support trolling	Engage in trolling
Specific behavior	Lurking	Use of SNS features to support trolling (upvotes, likes)	Direct trolling using comments and videos

If silent trolls exist, there must be factors that restrain or prompt overt trolling behaviors. We first test whether enjoyment of observing trolling is associated with supportive trolling, which then predicts overt trolling (see Figure 1 for the model). Then, we examine the moderating roles of fear of punishment and online toxic disinhibition in determining the associations (a) between enjoyment of observing trolling and supportive trolling and (b) between supportive and overt trolling behaviors, respectively. This study contributes to the literature by applying social cognitive theory to online trolling and by continuing to analyze the troll to audience relationship, which is necessary to understanding how these behaviors are engaged with and possibly replicated. It is our hope that this research will shed light on the scope of trolling behaviors.

Figure 1.

Hypothesized model.

Literature Review

Online Trolling

Online trolls are an integral part of the internet experience. While not desired by victims, the affordances of Web 2.0 computer-mediated communication, such as anonymity, invisibility, and asynchrony, allow average users to engage in non-conforming behavior in certain social environments (Suler, 2004). Despite a lack of consensus regarding the definition of online trolling, it is most often associated with online behavior that intentionally attempts to disrupt conversations (Herring et al., 2002), is provocative (Chen, 2018), and is marked by antisocial characteristics like aggression and deception (Hardaker, 2010). Unlike cyberbullying, trolling does not always involve power differences (Golf-Papez & Veer, 2017). Trolling also differs from flaming, which is more reactionary (Hmielowski et al., 2014), whereas trolling is intentional (Golf-Papez & Veer, 2017).

Research consistently finds that trolls are motivated by a sense of enjoyment of the reactions of their victims (Buckels et al., 2014; Cook et al., 2018). Trolls thrive off of the lulz, or the satisfaction at another’s suffering (Coleman, 2014). The vernacular lulz is nearly synonymous with the social scientific terms Schadenfreude (taking pleasure in another’s misfortunes; Dalakas et al., 2015) and sadism (pleasure in harming others; Buckels et al., 2013). Research confirms Schadenfreude’s (Brubaker et al., 2021) and sadism’s (Gylfason et al., 2021) association with trolling behaviors. Brubaker et al. (2021) found that a combined measure including Schadenfreude and sadism significantly mediated Dark Triad personality types (psychopathy, Machiavellianism, narcissism; Paulhus & Williams, 2002) and trolling behaviors.

Although much research conceptualizes trolling behaviors as direct attempts to provoke, disrupt, or deceive, little research examines media audiences who are exposed to trolling (e.g., Ubaradka & Khanganba, 2024). Research indicates that bystanders to online hate (Wachs & Wright, 2018) and trolling (Cheng et al., 2017) were likely to engage in each respective behavior. Moreover, Mao and Hu (2025) found that being a victim of trolling was associated with a greater likelihood of the victims engaging in reactive trolling. However, if such a definite division existed among all online users, online communities would simply be dichotomized into perpetrators and victims. The divisions are likely not so determined, leaving potential room for examining trolling bystanders and their motivations for consuming trolling content.

Online Bystanders

In social psychology, the bystander effect explains how witnesses to conflict or crimes react as a result of their perceived social environment. When bystanders perceive they are part of a larger group, they feel less sense of responsibility to intervene (Latanè & Darley, 1968). Brody and Vangelisti (2016) found this effect to be true in a study of bystanders’ reactions to cyberbullying: increased number of bystanders negatively predicted intervention. Wong-Lo et al. (2014) argued that one of the challenges presented to conceptualizations of bystanders in computer-mediated communication was how online environments produce large swaths of bystanders instantly and constantly as a result of the pervasiveness of cyber-aggression.

The lack of intervention against aggressive and antisocial online behaviors is problematic, yet there may be various reasons why bystanders refrain from corrective action. In addition to the reasons bystanders abdicate intervention in interpersonal settings, online users may feel like they lack the technical efficacy to effectively deal with aggressors and trolls (Wachs et al., 2019). Furthermore, users may feel if others are not intervening in a given online community, the behavior must be acceptable and even enjoyable. However, recent research has examined social media movements that promote civic moderation (Friess et al., 2021), finding that online counterspeech results in higher-quality posts and sublevel comments. Similarly, when Twitter users received negative feedback for racist comments, future racist behavior was reduced (Munger, 2017). Wachs et al. (2019) found that assertive coping (i.e., confrontation) and technical coping strategies (e.g., reporting, blocking) negatively moderated the relationship between being a bystander and perpetrator of online hate. In other words, just as social media present increased opportunities for trolling, social media also provide avenues of recourse that bystanders can choose to take.

A unique yet common phenomenon is presented when online bystanders enjoy trolling behaviors. Past media modalities have fostered enjoyment of aggressive and antisocial behaviors by audiences. But in traditional media, audiences have little power to participate or intervene in aggressive confrontations. Observing trolling behaviors is more akin to witnessing in-person discontent. Some research has incorporated both prosocial and antisocial forms of bystander intervention (Salmivalli, 1999; Salmivalli et al., 1996). Building on this research on antisocial forms of bystanders, Paull et al. (2012) argued that bystanders were usually actively or passively aligned with bullies or victims, or somewhere in between, rather than uninvolved outsiders. Bystanders could also be constructive or destructive. Paull et al. further declared that bystanders could transition between different roles as their knowledge of a scenario evolved. Specifically, they proposed a typology that categorized bystanders into 13 different versions, including facilitating bystanders (those that act as an audience to bullying and feel nothing wrong with their inactivity) and collaborating bystanders (those who actively joined in bullying by laughing at the victim or backing up the bully). Some research has utilized Paull et al.’s conception of active and passive bystanders in research analyzing workplace bullying (Ng et al., 2022) and has also incorporated Salmivalli et al.’s (1996) original bystander roles in research investigating cyberbullying (Song & Oh, 2018). However, we are unaware of any research that explicitly uses Paull et al.’s typology to understand trolling bystanders. We argue that facilitating and collaborating bystanders are manifest in the conceptualizations of silent inactive trolls (Brubaker et al., 2021), online users who enjoy watching suffering from afar, and silent supportive trolls, those who use social media features to support trolling, respectively. Recent literature has examined the phenomenon of passive trolling bystanders (Ubaradka & Khanganba, 2024) but relegated these users to only those who refrain from any type of behavior, closely aligned to who we consider silent inactive trolls. We will next discuss the basic dispositions and functions of inactive and supportive trolls and how they relate to more overt forms of trolling.

Enjoyment of Observing Trolling, Supportive Trolling, and Fear of Punishment

Brubaker et al. (2021) defined silent trolls as “online users who do not engage in trolling but still enjoy its presence in online discussion” (p. 10). Because they enjoy observing trolling, silent trolls have no intention to intervene. In fact, Brubaker et al. (2021) assert that these online lurkers are more than bystanders because of their personal involvement when consuming trolling behaviors. The authors envisioned silent trolls as possessing similar traits and motives as overt trolls but differing in their willingness to directly express opinions online.

Furthermore, Brubaker et al. (2021) suggested that silent trolls may participate in trolling behaviors in less outspoken ways, for example, upvoting or downvoting certain content. Indeed, computer-mediated communication features potentially allow for nuanced forms of trolling. For example, Phillips et al. (2021) found that users’ interest in the failures of others was positively associated with more clicks and more time on a debrief page detailing instances of failure than those interested in achievements. Ouwerkerk and Johnson (2016) found that social media users “hate-followed” other users they disliked so they could be alerted when something unfortunate happened to that person. We argue that where social media features are actively utilized (e.g., upvotes, likes) to engage in antisocial behavior online, behaviors resemble what can be termed silent supportive trolling. Supportive trolls go one step further than what we call silent inactive trolls. Thus, there is a difference between inactive trolling and supportive trolling, wherein supportive trolls enjoy observing trolling enough to click “like.” We hypothesize the following:

Hypothesis 1 (H1). Enjoyment of observed trolling will be positively associated with supportive trolling.

The hesitation to go from silent observation to active engagement may be for a variety of reasons. According to social cognitive theory (Bandura, 2001), people evaluate their desire to model observed behavior by whether the modeled behavior is rewarded or punished. Research confirms that deviant behavior is significantly influenced by perceived social rewards or punishments (Akers et al., 1979; Capece & Lanza-Kaduce, 2013). This effect may be unique in online environments. Research on context collapse (Marwick & Boyd, 2011) examines how people adapt their behavior in social media settings contingent on their perceived audience. Silent trolls may personally believe that trolling is an enjoyable behavior but may also be timid to perform these antisocial behaviors in front of potential familiar viewers and fear social punishment for doing so. These individuals we would consider silent inactive trolls, who dare not engage in any behavior resembling trolling. Moreover, based on Wachs and Wright’s (2018) finding that bystanders to online hate were also perpetrators, the silent troll may be a transitory period before the user finally engages in some form of observable trolling. This stage may involve overcoming the initial fear of punishment. As such, we hypothesize:

Hypothesis (H2). The relationship between enjoyment of observing trolling and supportive trolling will be positive for individuals with the lowest levels of fear of punishment. This relationship will decline in magnitudes as fear of punishment increases.

So far, we have examined what motivates silent observers to become supportive trolls as dependent on people’s fear of social punishment online. We now seek to understand what would motivate (or discourage) those who enjoy observing trolling and use social media features to support trolling to engage in overt trolling behaviors.

Overt Trolling and Online Disinhibition

Traditional trolling behaviors are actions that we can consider overt, or clearly observable, such as sending people to shock websites, sending controversial posts, or deliberately upsetting people online (Buckels et al., 2014; Craker & March, 2016). In fact, some trolls are self-aggrandizing and revel in their actions (Coleman, 2014). Yet people do not just wake up trolling others online. In line with social cognitive theory (Bandura, 2001), trolls would learn how to successfully model their tactics by observing the behaviors of other trolls. Social media provide a sort of playground for people to become proficient in trolling. Supportive trolling, using social media features such as likes and upvotes, could be a testing ground for people to assess the potential rewards and punishments before they engage in blatant trolling.¹ We predict the following:

Hypothesis 3 (H3). Supportive trolling will be positively associated with overt trolling.

It is possible that not all supportive trolls, who have overcome the initial fear of punishment, will engage in overt trolling. We argue that online disinhibition, or a shift in feelings of restraint online (Suler, 2004), potentially interacts with supportive trolling to influence whether one chooses to directly engage in trolling behaviors. Early research on social learning found that people felt greater disinhibition to engage in aggressive behavior when personal responsibility was diffused (Bandura et al., 1975). Suler (2004) explained how affordances of computer-mediated communication could enable individuals to feel less responsibility for antinormative behavior. For example, in online environments, anonymity could be used to avoid responsibility, and asynchronicity could delay immediate reactions to antisocial communication. The tendency for online affordances to induce antisocial behaviors that would not be performed in other interpersonal contexts is known as online toxic disinhibition (Suler, 2004).

Research indicates that online toxic disinhibition is associated with online aggression (Kurek et al., 2019) and trolling (Stuart & Scott, 2021). Furthermore, online toxic disinhibition often plays a moderating role in predicting online aggression. For instance, Wachs and Wright (2018) found that adolescent bystanders to online hate were more likely to also be perpetrators if they experienced increased disinhibition. Regarding trolling behavior, research has found that online disinhibition moderates the relationship between moral disengagement and online trolling (Wu et al., 2023). Mao and Hu (2025) observed that the mediated relationship between being a victim of trolling and reactive trolling through revenge motivation was stronger among those with higher online disinhibition. We may infer from these studies that trolling bystanders should experience greater online disinhibition because of their regular exposure to trolling. Potentially, then, online disinhibition strengthens the relationship between trolling exposure and more involved trolling behaviors. Supportive trolls who possess greater restraint should be more likely to refrain from overt trolling than those who feel greater online disinhibition. Therefore, we predict:

Hypothesis 4 (H4). The relationship between supportive trolling and overt trolling will be positive for individuals with the highest levels of online disinhibition. This relationship will decline in magnitude as online disinhibition decreases.

Method

Sample

This study analyzes data from a national online survey conducted in the United States between February 16 and March 21, 2024. The study was approved by the primary author’s university Institutional Review Board (IRB), and participants consented before completing the survey. We contracted the research company Dynata for data collection. Demographic quotas were applied for gender and age so that our sample would closely resemble the U.S. adult population (18 years and older) according to the U.S. Census Bureau. Respondents were 50.2% female. The median respondent was in the 45–54 age groups. To secure quality responses, we implemented two attention-check questions in the survey. Participants who failed any attention-check questions were not included in our final data set. Our final sample was N = 604.

Measures

Independent Variable

Enjoyment of Observed Trolling

The level to which respondents enjoy observing other users being trolled was adapted from Buckels et al. (2019) who asked participants what their level of enjoyment of trolling was when directed at five different targets, including: general public/strangers, other trolls, corporations, celebrities, and people you know offline. Participants first read a definition of online trolling stating, “Internet trolling is described as an intentional effort to use deception, aggression, or disruption to provoke other users and cause them anguish.” They were then asked to rate how much they enjoy observing each target being trolled on a five-point scale from 1 do not enjoy at all to 5 very much enjoy. For participants who had not seen certain targets being trolled, they were asked to indicate what their enjoyment would be. Items were averaged (α = .89; M = 1.99, SD = 0.97).

Mediator

Supportive Trolling

Supportive trolling was defined as online users’ utilization of “likes” or “upvotes” to express enjoyment of trolling. To measure supportive trolling, we adapted seven items from the Global Assessment of Facebook Trolling (Craker & March, 2016) but reworded statements to include the actions of “like” or “upvote.” Sample items included, “I like/upvote disturbing or controversial online posts someone shared or sent,” “I avoid liking/upvoting posts that cause controversy or stir up trouble (reverse coded).” Each item was measured on a five-point scale from 1 never to 5 always. Items were averaged (α = .75; M = 1.99, SD = 0.72).

Dependent Variable

Overt Trolling

To measure overt trolling behaviors, we adapted seven items from the Global Assessment of Facebook Trolling (Craker & March, 2016). Sample items included “I share or send disturbing or controversial online posts for fun” and “I avoid causing controversy or stirring up trouble online (reverse coded).” Each item was measured on a five-point scale from 1 never to 5 always. Items were averaged (α = .76; M = 1.79, SD = 0.66).

Moderators

Fear of Punishment

We measured fear of punishment by adapting four items from Caprara et al.’s (1992) fear of punishment scale. Items included, “The thought of being punished for my mistakes online is a source of anguish for me,” “It’s worth telling lies online to avoid the consequences of your actions,” “I sometimes think with fear about the consequences of what I’ve done or said online,” and “I’m afraid that people might get to know about some of the things I’ve done online.” Each item was measured from 1 strongly disagree to 5 strongly agree. Items were averaged (α = .77; M = 2.07, SD = 0.87).

Online Disinhibition

We adapted Udris’s (2014) four-item scale for toxic online disinhibition. Sample items included, “I don’t mind writing insulting things about others online, because it’s anonymous” and “There are no rules online therefore you can do whatever you want.” Each item was measured from 1 definitely do not believe to 5 definitely do believe. Items were combined and averaged (α = .68; M = 1.85, SD = 0.78).

Control Variables

We controlled gender² and age. Research indicates that males and females respond to and engage in online antisocial behavior differently (Kurek et al., 2019). In addition, trolling research indicates that younger (vs. older) people are more likely to engage in trolling (Buckels et al., 2019; Craker & March, 2016). Next, because research indicates significant relationships between Dark Tetrad personality types and trolling behavior (Buckels et al., 2019) and between Schadenfreude and trolling (Brubaker et al., 2021), we controlled for overall Dark Tetrad personality types (i.e., the short Dark Tetrad; Paulhus et al., 2021; α = .91; M = 2.53, SD = 0.61)³ and trait Schadenfreude (James et al., 2014; α = .87; M = 4.27, SD = 1.00). Finally, because our survey relies on self-report, we controlled for social desirability using Hays et al. (1989) Socially Desirable Response Set-5 (α = .60; M = 3.68, SD = 0.66).

Data Analysis

To test our dual moderated mediation model (Figure 1), we used PROCESS macro Model 21 (Hayes, 2018) with 5,000 bootstrap samples and 95% confidence intervals. All interaction variables were mean-centered.

Results

We first ran bivariate correlations for all main variables and covariates (Table 2). To test for H1, we conducted an ordinary least squares regression, regressing supportive trolling on enjoyment of observing trolling while controlling for several covariates (Table 3, Column 1). H1 was supported; enjoyment of observing trolling was positively associated with supportive trolling (b = .23, SE = .03, p < .001).

Table 2.

Bivariate Correlations.

Variables	1	2	3	4	5	6	7	8	9
1. Overt trolling
2. Support. trolling	.78**
3. Enjoyment	.56**	.57**
4. Fear of punish.	.41**	.43**	.42**
5. Disinhibition	.52**	.48**	.44**	.35**
6. Dark Tetrad	.58**	.55**	.57**	.44**	.42**
7. Schadenfreude	.25**	.26**	.34**	.16**	.26**	.38**
8. Soc. desirability	−.43**	−.41**	−.44**	−.35**	−.40**	−.46**	−.44**
9. Age	−.18**	−.21**	−.30**	−.28**	−.07	−.23**	.01	.12*
10. Male	.15**	.15**	.15**	.03	.19**	.24**	.14**	−.12*	.27**

Note. Male (0 = female, 1 = male).

p < .01. **p < .001.

Table 3.

Predicting Supportive and Overt Trolling.

	Supportive trolling		Overt trolling
	b (SE)	b (SE)	b (SE)	b (SE)
Constant	1.97 (.03)***	1.95 (.03)***	1.80 (.02)***	1.77 (.02)***
Enjoyment	0.23 (.03)***	0.22 (.03)***	0.05 (.02)*	0.05 (.02)*
Supportive trolling	–	–	0.52 (.03)***	0.48 (.03)***
Fear of punishment	0.12 (.03)***	0.10 (.03)**	0.004 (.02)	−0.01 (.02)
Enjoy x Fear	–	0.06 (.02)**	–	–
Disinhibition	–	–	0.11 (.03)***	0.08 (.03)**
Supp. x Disinhib.	–	–	–	0.10 (.02)***
Dark Tetrad	0.31 (.05)***	0.30 (.05)***	0.18 (.04)***	0.17 (.04)***
Schadenfreude	−0.01 (.03)	−0.00 (.03)	−0.04 (.02)#	−0.03 (.02)
Social des.	−0.11 (.04)*	−0.11 (.04)**	−0.06 (.03)*	−0.08 (.03)**
Age	−0.002 (.02)	−0.003 (.02)	0.01 (.01)	0.01 (.01)
Male	0.04 (.05)	0.04 (.05)	−0.03 (.04)	−0.02 (.03)
R Square	.43	.43	.66	.67
df	(7, 594)	(8, 593)	(9, 592)	(10, 591)

Note. Continuous variables were mean-centered. Male (0 = female, 1 = male).

***

p < .001. **p < .01. *p < .05. #p < .10.

Next, we analyzed H2, which predicted that the relationship between enjoyment of observing trolling and supportive trolling would be positive for individuals with the lowest levels of fear of punishment. To examine this, we included the interaction term between enjoyment of observing trolling and fear of punishment (Table 3, Column 2) in the model. Results indicated a significant but positive interaction, contrary to expectations (b = .06, SE = .02, p = .009). To further understand this interaction, we ran a simple moderation analysis (Table 4) by setting the value of the moderator to one standard deviation below the mean, at the mean, and one standard deviation above the mean (Hayes, 2018). Enjoyment of observing trolling has a positive relationship with supportive trolling among individuals with low (b = .17, SE = .04, p < .001), medium (b = .22, SE = .03, p < .001) and high levels of fear of punishment (b = .28, SE = .03, p < .001); the positive relationship becomes stronger as fear of punishment increases. H2 was not supported.

Table 4.

Conditional Relationship Between Enjoyment of Observing Trolling and Supportive Trolling at Values of Fear of Punishment.

Fear of punishment	Point estimate	95% CI
−.87 (−1SD)	.17 (.04)	[.09, .25]
.00 (M)	.22 (.03)	[.16, .28]
.87 (+1SD)	.28 (.03)	[.21, .34]

Next, H3 predicted that supportive trolling would be positively associated with overt trolling. We ran another ordinary least squares regression, regressing overt trolling on supportive trolling while controlling for enjoyment of observing trolling, fear of punishment, online disinhibition, and the covariates (Table 3, Column 3). Overt trolling was positively associated with supportive trolling (b = .52, SE = .03, p < .001) and online disinhibition (b = .11, SE = .03, p < .001).

We then added the interaction between supportive trolling and online disinhibition to the model (Table 3, Column 4) to test for H4, which predicted that the relationship between supportive trolling and overt trolling would be positive for individuals with the highest level of online disinhibition, and this relationship would decline as online disinhibition levels declined. Analysis produced a positive and significant effect (b = .10, SE = .02, p < .001). We then probed this interaction with another simple moderation analysis (Table 5). Supportive trolling has a positive relationship with overt trolling among individuals with low (b = .41, SE = .04, p < .001), medium (b = .48, SE = .03, p < .001), and high levels of online disinhibition (b = .56, SE = .03, p < .001); the positive relationship becomes stronger as online disinhibition increases.⁴ Overall, we found support for the dual moderated mediation when fear of punishment is the first-stage moderator and online disinhibition is the second-stage moderator (b = .01, SE = .004, CI = [.000, .015]).

Table 5.

Conditional Relationship Between Supportive Trolling and Overt Trolling at Values of Disinhibition.

Disinhibition	Point estimate	95% CI
−.78 (−1SD)	.41 (.04)	[.33, .48]
.00 (M)	.48 (.03)	[.42, .54]
.78 (+1SD)	.56 (.03)	[.50, .62]

Discussion

Based on the theoretical framework of social cognitive theory (Bandura, 2001), the interpersonal bystander effect (Latanè & Darley, 1968) and proposed nuanced forms of trolling (Brubaker et al., 2021), we analyzed data from an online survey to examine how enjoyment of observing trolling behaviors online could predict more direct forms of trolling: supportive and overt trolling behaviors. We hypothesized that social cognitive motivators (Bandura, 2001), such as fear of punishment and disinhibition, would encourage or dissuade people from going from mere lurkers to supportive and overt trolls. We also considered how social media features such as likes and upvotes enable people to support online trolling without making direct comments to others. Our findings first showed that enjoyment of observing trolling was positively associated with supportive trolling. This relationship was moderated by fear of punishment, but in the opposite direction of what was expected; as fear increased, the relationship between enjoyment and supportive trolling became stronger. Supportive trolling then was positively associated with overt trolling, and this relationship was strongest for people with the highest levels of online disinhibition, thus confirming our hypothesis. These results accentuate how perceptions of social environments and one’s immersion in online communities affect people’s attitudes toward and engagement in trolling. We discuss each finding in further detail below.

First, our findings confirmed that enjoyment of trolling was associated with supportive trolling, which in turn was positively associated with overt trolling. The results confirm our assumptions that supportive trolling could be a critical steppingstone into overt trolling among people who really enjoy observing trolling. In online communities, individuals can easily express their support for trolling behaviors with just one click of a like or upvote button. We speculate if self-efficacy is developed through repeated acts of supportive trolling, per social cognitive theory (Bandura, 2001), these individuals will try overt trolling themselves. However, future research is needed to examine the mechanism behind this mediation relationship and confirm whether this is a causal effect.

Next, the results disconfirmed our first moderation hypothesis. Although fear of punishment significantly moderated the relationship between enjoyment of observing trolling and supportive trolling, this relationship was surprisingly strongest for people who had the highest levels of fear of punishment and weakest for those with the lowest fear of punishment. Although this finding seems to be at odds with what social cognitive theory suggests, it may be that unlike people who refrain from any form of trolling and thus have nothing to fear, individuals who engage in supportive trolling have something to fear retrospectively. It could be that they enjoy observing trolling so much that they are willing to do something that may result in mild retribution. While these individuals are willing to like trolling content on social media, they are aware that their behavior will likely be shown on their social networks’ feeds, and they may possibly be negatively evaluated and unfriended by others. Prior studies have shown that contents that are offensive or “piss people off” (John & Dvir-Gvirsman, 2015; Vraga et al., 2015, p. 283) can result in unfriending and social media fights. People who positively react to trolling content are associating themselves with offensive and annoying content and may worry what their friends and associates will indeed see what they have liked. Indeed, prior scholarship discusses the importance of association and visibility affordances of social media that by liking certain contents, users are associating themselves with the contents, which may be visible to other users they are connected with (Treem & Leonardi, 2013). Furthermore, these engagements such as liking are persistent, meaning that other users can also access them at a later time (Kim & Ellison, 2022).

Future research should examine whether supportive trolls ever remove likes or upvotes, or whether their fear stems from activities that they still allow to be visible and why they would do so. Also, we believe personal prosocial values could weaken the link between enjoyment of observing trolling and supportive trolling. Future research could analyze how silent trolls evaluate their feelings of enjoyment of observing trolling and how those conflict with personal values. It will also be worthwhile to distinguish between people who actively look for trolling content to observe and those who are inadvertently exposed to trolling and whether prosocial values are higher among the latter.

In the context of social cognitive theory, research could also assess whether supportive and overt trolls perceive and anticipate higher rewards for their behavior that outweigh punishment. According to our results (and assuming a causal path), once a person reaches supportive trolling, overt trolling seems likely, confirming Wachs and Wright’s (2018) findings about exposure to online hate and the positive association with enactment of such behavior. We predicted that the relationship between supportive and overt trolling would be strongest among people who felt the highest levels of online disinhibition. While this was true, we also found a significant, positive relationship at medium and low levels of online disinhibition. Other research has found that the mediated relationship between trolling victimization and reactive trolling through revenge motivation is moderated by online disinhibition (Mao & Hu, 2025). Future research should examine the relationship between being a trolling victim and supportive trolling to see if victims also follow a similar trajectory before finally engaging in trolling behaviors.

On a positive note, average levels of enjoyment of observing trolling and engaging in supportive and overt trolling were relatively low among our sample, with means being below their respective midpoints. While this is a somewhat reassuring finding, among those that really enjoy observing trolling, however, trolling behaviors are highly likely to accompany their positive attitudes. This finding has strong implications for the features of social media (e.g., like, upvote, share, comment). Essentially, if people regularly see trolling behavior, social media may facilitate their gradual transformation into a troll, assuming that the relationship is causal. The other possibility is that trolls intentionally frequent online discussion platforms because they know where they can get their “kicks.” Experimental research is needed to uncover this relationship further and determine whether social media features make it more likely to engage in antisocial forms of behavior than in interpersonal settings. Simultaneously, research can examine how social desirability or prosocial attitudes can weaken the positive links between enjoyment, supportive trolling, and overt trolling to (1) further explore the existence of silent trolls and (2) so that antitrolling entities can learn how to break these links and salvage online discussions from incessant trolling.

Recent research has studied the phenomenon of bystander trolling (Ubaradka & Khanganba, 2024), but did not incorporate behaviors relating to social media features (e.g., likes, upvotes). Thus, our study helps to unveil the mechanism through which bystanders potentially go on to engage in more overt forms of trolling. We also acknowledge that Bandura et al.’s (1975) research on disinhibition and deviant behaviors conceptualized diffused personal responsibility in terms of group collective action. In our study, we have understood decreased personal responsibility through the affordances of social media features. While Brody and Vangelisti (2016) found that those who perceived a high number of bystanders to cyberbullying had less intention to intervene, future research should examine whether perceptions of shared responsibility or social media features have a greater effect on decreasing personal responsibility when engaging in online antisocial behaviors.

Limitations and Future Research

Our study includes several limitations. The first is that we used self-report survey methods. For this reason, we controlled for social desirability in our analysis. Next, items we used to assess supportive trolling asked participants their willingness to “like/upvote” certain trolling behaviors. While similar, these behaviors are not interchangeable. The like feature is attached to a user’s profile and publicly visible on platforms like Facebook, whereas visibility of upvote clickers varies by platforms and communities (e.g., not visible on Reddit, visible on Quora). This nuance potentially confounded our participants’ interpretation of the scale such that some feared “liking” trolling content but not upvoting trolling content. Future studies will need to analyze how inactive and supportive trolls operate in different online spaces and should develop measurements that can be used more broadly or be tailored to specific platforms.

In addition, our surprising findings regarding fear of punishment might be the result of the scale we used, which largely assessed fear of past actions. In addition to the retrospective nature of our fear of punishment measurement, the present scale reflects overt antisocial behaviors. Future research should adapt items to consider both anticipatory fear and less conspicuous forms of trolling.

Another limitation of our study is that our survey was administered to a sample with age and gender quotas that resembles the U.S. adult population. Currently, social media users tend to be younger, with sites like Facebook, Instagram, and TikTok used more among women than men (Pew Research Center, 2024). However, we believe that it is essential to understand how different age groups perceive what their role in response to observing trolling is. Some age groups are more vulnerable to being affected by antisocial behavior than others, and future work should be conducted to examine how these vulnerable groups (e.g., older people, young adults) internalize trolling.

A final limitation is that our survey design cannot prove causality. We acknowledge that it could very well be that those actively engaged in overt trolling are also more likely to engage in supportive trolling and enjoy observing others troll. While our hypotheses were theoretically driven, we also tested our model for reverse causality. A moderated mediation test revealed no indirect relationship between overt trolling and enjoyment of observing trolling through supportive trolling when online disinhibition moderates the first link and fear of punishment moderates the second link (b = .01, SE = .01, CI = [−.002, .016]), lending some support to our model. Thus, we argue that our hypothesized model is plausible. Still, future research should test our hypotheses using experimental designs. For example, after exposure to fictitious online trolling behaviors, researchers can measure participants’ enjoyment of trolling activity and provide opportunities to engage with the stimuli through social media features.

Conclusion

This study holds important implications for trolling research and coping strategies when confronted with trolling behaviors. Our results suggest that psychological disinhibition may be the key to breaking the link between supportive and overt trolling. Through this understanding, individuals and communities can make concerted efforts to increase self-awareness of traditional and nuanced forms of online trolling and take preventive steps to stop it from spreading. Furthermore, preventive steps need not involve attacking online trolls. Research shows that those who engage in online counterspeech seek ways to moderate uncivil discussion by rewarding positive comments and avoiding replies to aggressive comments, thereby manipulating social media algorithms to push down uncivil content (Buerger, 2021).

The affordances and features of online communication make this environment a unique place for antisocial behaviors to be committed and observed. Our study examined how enjoyment of observing trolling is associated with supportive trolling, which is accompanied by overt trolling. We then examined how social cognitive factors like fear of punishment and disinhibition moderate this relationship. It is our hope that as knowledge about these findings increases, everyday internet users can make better decisions in their own online behavior.

Supplemental Material

sj-docx-1-sms-10.1177_20563051251320437 – Supplemental material for How Do Silent Trolls Become Overt Trolls? Fear of Punishment and Online Disinhibition Moderate the Trolling Path

Supplemental material, sj-docx-1-sms-10.1177_20563051251320437 for How Do Silent Trolls Become Overt Trolls? Fear of Punishment and Online Disinhibition Moderate the Trolling Path by Daniel Montez and Dam Hee Kim in Social Media + Society

Footnotes

Declaration of Conflicting Interests

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

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the Department of Communication at the University of Arizona and the School of Media and Communication at Korea University.

Data Availability Statement

The data that support the findings of this study are available on request from the corresponding author.

ORCID iD

Daniel Montez

Supplemental Material

Supplemental material for this article is available online.

Notes

Author Biographies

Daniel Montez (PhD, University of Arizona) is an adjunct lecturer in the Department of Communication at the University of Arizona. His research interests focus on political communication, (in)civility in political discussion, online trolling, and media effects.

Dam Hee Kim (PhD, University of Michigan) is an Associate Professor in the School of Media and Communication at Korea University. With her cross-disciplinary background in communication and business, her research focuses on the social, political, and business implications of individuals’ engagement with news and entertainment content in the evolving media environment.

References

Akers

R. L.

Krohn

M. D.

Lanza-Kaduce

Radosevich

(1979). Social learning and deviant behavior: A specific test of a general theory. American Sociological Review, 44(4), 187–214. https://doi.org/10.2307/2095052

Bandura

(2001). Social cognitive theory of mass communication. Media Psychology, 3(3), 265–299. https://doi.org/10.1207/S1532785XMEP0303_03

Bandura

Underwood

Fromson

M. E.

(1975). Disinhibition of aggression through diffusion of responsibility and dehumanization of victims. Journal of Research in Personality, 9(4), 253–269. https://doi.org/10.1016/0092-6566(75)90001-X

Brody

Vangelisti

A. L.

(2016). Bystander intervention in cyberbullying. Communication Monographs, 83(1), 94–119. https://doi.org/10.1080/03637751.2015.1044256

Brubaker

P. J.

Montez

Church

S. H.

(2021). The power of schadenfreude: Predicting behaviors and perceptions of trolling among Reddit users. Social Media + Society, 7(2), 1–13. https://doi.org/10.1177%2F20563051211021382

Buckels

E. E.

Jones

D. N.

Paulhus

D. L.

(2013). Behavioral confirmation of everyday sadism. Psychological Science, 24(11), 2201–2209. https://doi.org/10.1177/0956797613490749

Buckels

E. E.

Trapnell

P. D.

Andjelovic

Paulhus

D. L.

(2019). Internet trolling and everyday sadism: Parallel effects on pain perception and moral judgment. Journal of Personality, 87(2), 328–340. https://doi.org/10.1111/jopy.12393

Buckels

E. E.

Trapnell

P. D.

Paulhus

D. L.

(2014). Trolls just want to have fun. Personality and Individual Differences, 67, 97–102. https://doi.org/10.1016/j.paid.2014.01.016

Buerger

(2021). #iamhere: Collective counterspeech and the quest to improve online discourse. Social Media + Society, 7(4), 20563051211063843. https://doi.org/10.1177%2F20563051211063843

10.

Capece

Lanza-Kaduce

(2013). Binge drinking among college students: A partial test of Akers’ social structure-social learning theory. American Journal of Criminal Justice, 38, 503–519. https://doi.org/10.1007/s12103-013-9208-4

11.

Caprara

G. V.

Manzi

Perugini

(1992). Investigating guilt in relation to emotionality and aggression. Personality and Individual Differences, 13(5), 519–532. https://doi.org/10.1016/0191-8869(92)90193-S

12.

Chen

(2018). “Being a butt while on the internet”: Perceptions of what is and isn’t internet trolling. Proceedings of the Association for Information Science and Technology, 55(1), 76–85. https://doi.org/10.1002/pra2.2018.14505501009

13.

Cheng

Bernstein

Danescu-Niculescu-Mizil

Leskovec

(2017, February). Anyone can become a troll: Causes of trolling behavior in online discussions. In Proceedings of the 2017 ACM Conference on Computer Supported Cooperative Work and Social Computing (pp. 1217–1230). Association for Computing Machinery. https://doi.org/10.1145/2998181.2998213

14.

Coleman

(2014). Hacker, hoaxer, whistleblower, spy: The many faces of anonymous. Verso Books.

15.

Cook

Schaafsma

Antheunis

(2018). Under the bridge: An in-depth examination of online trolling in the gaming context. New Media & Society, 20(9), 3323–3340. https://doi.org/10.1177%2F1461444817748578

16.

Craker

March

. (2016). The dark side of Facebook®: The Dark Tetrad, negative social potency, and trolling behaviours. Personality and Individual Differences, 102, 79–84. https://doi.org/10.1016/j.paid.2016.06.043

17.

Dalakas

Phillips Melancon

Sreboth

(2015). A qualitative inquiry on schadenfreude by sport fans. Journal of Sport Behavior, 38(2), 161–179.

18.

Friess

Ziegele

Heinbach

(2021). Collective civic moderation for deliberation? Exploring the links between citizens’ organized engagement in comment sections and the deliberative quality of online discussions. Political Communication, 38(5), 624–646. https://doi.org/10.1080/10584609.2020.1830322

19.

Gallegos

(2021, November 30). U.S. millennials are most likely to engage in trolling, study finds. The Avast Foundation.

20.

Goffman

(1959). The presentation of self in everyday life. Doubleday.

21.

Golf-Papez

Veer

(2017). Don’t feed the trolling: Rethinking how online trolling is being defined and combated. Journal of Marketing Management, 33(15–16), 1336–1354. https://doi.org/10.1080/0267257X.2017.1383298

22.

Gylfason

H. F.

Sveinsdottir

A. H.

Vésteinsdóttir

Sigurvinsdottir

(2021). Haters gonna hate, trolls gonna troll: The personality profile of a Facebook troll. International Journal of Environmental Research and Public Health, 18(11), 5722. https://doi.org/10.3390/ijerph18115722

23.

Hardaker

(2010). Trolling in asynchronous computer-mediated communication: From user discussions to academic definitions. Journal of Politeness Research: Language, Behaviour, Culture, 6(2), 215–242. https://doi.org/10.1515/jplr.2010.011

24.

Hayes

A. F.

(2018). Introduction to mediation, moderation, and conditional process analysis: A regression-based approach (2nd ed.). Guilford Press.

25.

Hays

R. D.

Hayashi

Stewart

A. L.

(1989). A five-item measure of socially desirable response set. Educational and Psychological Measurement, 49(3), 629–636. https://doi.org/10.1177/001316448904900315

26.

Herring

Job-Sluder

Scheckler

Barab

(2002). Searching for safety online: Managing “trolling” in a feminist forum. The Information Society, 18(5), 371–384. https://doi.org/10.1080/01972240290108186

27.

Hmielowski

J. D.

Hutchens

M. J.

Cicchirillo

V. J.

(2014). Living in an age of online incivility: Examining the conditional indirect effects of online discussion on political flaming. Information, Communication & Society, 17(10), 1196–1211. https://doi.org/10.1080/1369118X.2014.899609

28.

Hogan

(2010). The presentation of self in the age of social media: Distinguishing performances and exhibitions online. Bulletin of Science, Technology & Society, 30(6), 377–386. https://doi.org/10.1177%2F0270467610385893

29.

James

Kavanagh

P. S.

Jonason

P. K.

Chonody

J. M.

Scrutton

H. E.

(2014). The Dark Triad, schadenfreude, and sensational interests: Dark personalities, dark emotions, and dark behaviors. Personality and Individual Differences, 68, 211–216. https://doi.org/10.1016/j.paid.2014.04.020

30.

John

N. A.

Dvir-Gvirsman

(2015). “I don’t like you any more”: Facebook unfriending by Israelis during the Israel–Gaza conflict of 2014. Journal of Communication, 65(6), 953–974. https://doi.org/10.1111/jcom.12188

31.

Kim

D. H.

Ellison

N. B.

(2022). From observation on social media to offline political participation: The social media affordances approach. New Media & Society, 24(12), 2614–2634. https://doi.org/10.1177/1461444821998346

32.

Kurek

Jose

P. E.

Stuart

(2019). ‘I did it for the LULZ’: How the dark personality predicts online disinhibition and aggressive online behavior in adolescence. Computers in Human Behavior, 98, 31–40. https://doi.org/10.1016/j.chb.2019.03.027

33.

Latanè

Darley

J. M.

(1968). Group inhibition of bystander intervention in emergencies. Journal of Personality and Social Psychology, 10(3), 215–221. https://doi.org/10.1037/h0026570

34.

Mao

(2025). From trolling victimization to reactive trolling: Moderated mediation effects of online disinhibition and motivations. Social Science Computer Review, 43, 27–47. https://doi.org/10.1177/08944393241250013

35.

Marrington

J. Z.

March

Murray

Jeffries

Machin

March

. (2023). An exploration of trolling behaviours in Australian adolescents: An online survey. PLOS ONE, 18(4), Article e0284378. https://doi.org/10.1371/journal.pone.0284378

36.

Marwick

A. E.

Boyd

(2011). I tweet honestly, I tweet passionately: Twitter users, context collapse, and the imagined audience. New Media & Society, 13(1), 114–133. https://doi.org/10.1177/1461444810365313

37.

Munger

(2017). Tweetment effects on the Tweeted: Experimentally reducing racist harassment. Political Behavior, 39(3), 629–649. https://doi.org/10.1007/s11109-016-9373-5

38.

Niven

Notelaers

(2022). Does bystander behavior make a difference? How passive and active bystanders in the group moderate the effects of bullying exposure. Journal of Occupational Health Psychology, 27(1), 119–135. https://doi.org/10.1037/ocp0000296

39.

Ouwerkerk

J. W.

Johnson

B. K.

(2016). Motives for online friending and following: The dark side of social network site connections. Social Media + Society, 2(3), 1–13. https://doi.org/10.1177/2056305116664219

40.

Paulhus

D. L.

Buckels

E. E.

Trapnell

P. D.

Jones

D. N.

(2021). Screening for dark personalities. European Journal of Psychological Assessment, 37(3), 208–222. https://doi.org/10.1027/1015-5759/a000602

41.

Paulhus

D. L.

Williams

K. M.

(2002). The dark triad of personality: Narcissism, machiavellianism, and psychopathy. Journal of Research in Personality, 36(6), 556–563. https://doi.org/10.1016/S0092-6566(02)00505-6

42.

Paull

Omari

Standen

(2012). When is a bystander not a bystander? A typology of the roles of bystanders in workplace bullying. Asia Pacific Journal of Human Resources, 50(3), 351–366. https://doi.org/10.1111/j.1744-7941.2012.00027.x

43.

Pew Research Center. (2024, January 31). Social media fact sheet. Pew Research Center. https://www.pewresearch.org/internet/fact-sheet/social-media/?tabId=tab-0ec23460-3241-4a1f-89bc-0c27fb641936

44.

Phillips

J. G.

Landhuis

C. E.

Wood

(2021). Appreciation of achievement and a behavioural preference to view failure: Schadenfreude online. Personality and Individual Differences, 172, 110597. https://doi.org/10.1016/j.paid.2020.110597

45.

Salmivalli

(1999). Participant role approach to school bullying: Implications for intervention. Journal of Adolescence, 22, 453–459. https://doi.org/10.1006/jado.1999.0239

46.

Salmivalli

Lagerspetz

Björkqvist

Österman

Kaukiainen

(1996). Bullying as a group process: Participant roles and their relations to social status within the group. Aggressive Behavior, 22(1), 1–15. https://doi.org/10.1002/(SICI)1098-2337(1996)22:1%3C1::AID-AB1%3E3.0.CO;2-T

47.

Song

(2018). Factors influencing bystanders’ behavioral reactions in cyberbullying situations. Computers in Human Behavior, 78, 273–282. https://doi.org/10.1016/j.chb.2017.10.008

48.

Stuart

Scott

(2021). The measure of online disinhibition (MOD): Assessing perceptions of reductions in restraint in the online environment. Computers in Human Behavior, 114, 106534. https://doi.org/10.1016/j.chb.2020.106534

49.

Suler

(2004). The online disinhibition effect. Cyberpsychology & Behavior, 7(3), 321–326. https://doi.org/10.1089/1094931041291295

50.

Treem

J. W.

Leonardi

P. M.

(2013). Social media use in organizations: Exploring the affordances of visibility, editability, persistence, and association. Annals of the International Communication Association, 36(1), 143–189. https://doi.org/10.1080/23808985.2013.11679130

51.

Ubaradka

Khanganba

S. P.

(2024). The differential effect of psychopathy on active and bystander trolling behaviors: The role of dark tetrad traits and lower agreeableness. Scientific Reports, 14(1), 9905. https://doi.org/10.1038/s41598-024-60203-6

52.

Udris

(2014). Cyberbullying among high school students in Japan: Development and validation of the Online Disinhibition Scale. Computers in Human Behavior, 41, 253–261. https://doi.org/10.1016/j.chb.2014.09.036

53.

Vogels

E. A.

(2021, January 13). The state of online harassment. Pew Research Center. https://www.pewresearch.org/internet/2021/01/13/the-state-of-online-harassment/

54.

Vraga

E. K.

Thorson

Kligler-Vilenchik

Gee

(2015). How individual sensitivities to disagreement shape youth political expression on Facebook. Computers in Human Behavior, 45, 281–289. https://doi.org/10.1016/j.chb.2014.12.025

55.

Wachs

Wright

M. F.

(2018). Associations between bystanders and perpetrators of online hate: The moderating role of toxic online disinhibition. International Journal of Environmental Research and Public Health, 15(9), 2030. https://doi.org/10.3390/ijerph15092030

56.

Wachs

Wright

M. F.

Sittichai

Singh

Biswal

Kim

E. M.

Yang

Gámez-Guadix

Almendros

Flora

Daskalou

Maziridou

(2019). Associations between witnessing and perpetrating online hate in eight countries: The buffering effects of problem-focused coping. International Journal of Environmental Research and Public Health, 16(20), 3992. https://doi.org/10.3390/ijerph16203992

57.

Wong-Lo

Bullock

L. M.

(2014). Digital metamorphosis: Examination of the bystander culture in cyberbullying. Aggression and Violent Behavior, 19(4), 418–422. https://doi.org/10.1016/j.avb.2014.06.007

58.

Xiao

Zhou

Liu

(2023). Why individuals with psychopathy and moral disengagement are more likely to engage in online trolling? The online disinhibition effect. Journal of Psychopathology and Behavioral Assessment, 45(2), 322–332. https://doi.org/10.1007/s10862-023-10028-w

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.03 MB