Abstract
Uncivil online communication is a widely problematized cultural by-product of computer-mediated communication that, however, remains theoretically underexplained. While previous research shows that personal tendency for uncivil communication is partially influenced by individuals’ personality and empathy skills, the factors of inter-individual variation remain largely unknown. The present study examined individuals’ emotion regulation skills as a possible predictor of uncivil communication. Online survey respondents (N = 215) reported if they had engaged in uncivil communication and filled in scales measuring emotion regulation difficulties, use of different emotion regulation strategies, and various individual traits. The results show that emotion regulation difficulties were associated with high levels of online disinhibition. This, in turn, was associated with reports of uncivil communication. The mediation effect was observed even when controlling for personality and empathy. The results suggest that individuals’ emotion regulation difficulties may be an underlying psychological factor behind harmful online communication. These findings call for research and development of means to support emotion regulation in online interactions.
Keywords: Uncivil communication, Online disinhibition, Computer-mediated communication, Social media, Online discussion, Digital emotion regulation
Subject terms: Human behaviour, Information technology
Introduction and theoretical framework
While the internet has revolutionized interpersonal communication, the computer-mediated and often text-based form of communication has introduced new social and cultural problems. One persistent issue on many online platforms is the prevalence of expressions of hatred, personal attacks, and other forms of uncivil communication1. Uncivil communication can have various detrimental consequences, such as hindering constructive discourse2, polarizing opinions in the society3and lowering the perceived quality of online platforms4. Previous research has shown that specific individual traits such as narcissism and low empathy are associated with online incivility5–7. However, we still lack comprehensive understanding of the individual-level factors that predict engagement in such behavior at large.
We suggest that a key gap in the literature relates to individual differences in emotion regulation processes. Experimental studies have demonstrated that users’ negative emotions are one of the causal mechanisms that trigger uncivil online behavior8,9. Following this, it seems likely that individuals’ online behavior could be influenced by how they regulate their emotions but to the best of our knowledge the associations between individuals’ emotion regulation skills and uncivil communication have not been studied. Emotion regulation seems to be more difficult in textual computer-mediated communication as compared to face-to-face situations. Research on media studies implies that the attenuation of emotion regulation online is associated with factors identified in the online disinhibition effect10. For example, anonymity, invisibility, asynchronicity, and minimization of authority in online communication may be associated with the phenomenon. Emotion regulation is theoretically linked to online disinhibition because poor regulation can diminish self-control in digital spaces, where the typical social norms regulating the behavior are weakened.
In this study, we conducted an online survey to examine whether individual differences in emotion regulation (i.e., perceived difficulties in emotion regulation and use of different emotion regulation strategies) are associated with uncivil online communication. It is noted that we mostly focused on incivility forms related to speech-based norms, and only indirectly addressed violations of the inclusion-based norms11. The findings provide new information about the associations between uncivil online behavior and emotion regulation and have practical significance in the development of interventions that could support emotion regulation in online communication.
We will next present the theoretical and empirical background of the current study, describing how the concepts measured in the study are related to uncivil online behavior according to prior literature.
Theoretical background
Uncivil Communication
Uncivil expressions, toxic language, and antisocial behavior1,12 are common on social media platforms. Coe et al.1reported that approximately one-fifth of examined online news comments contained incivility. Surveying social media interactions of UK parliament members, Akhtar and Morrison13 found that every respondent reported having been targets of online trolling as one form of antisocial behavior. These findings suggest that many behaviors that would be considered inappropriate in face-to-face interactions are common in online contexts.
While it is accepted that online incivility is common and has wide societal repercussions, the concept of incivility is difficult to define1,6. The concept can be divided into personal-level incivility, which refers to impoliteness and rude choices of words, and public-level incivility, which refers to a lack of deliberativeness and reciprocity, hence threatening democracy11,14. The former term can also be referred to as uncivil discourse and the later as intolerant discoursewhich may not always be perceived as uncivil, but individuals or groups for example minorities, are attacked in ways that is morally questionable and threatens democracy15. In this study, the use of the concept of incivility focuses more on acts that violate interpersonal politeness norms, which may include for example name-calling, aspersion, vulgarity, or use of pejoratives1. However, there is no objective definition of what is or is not uncivil communication. Communication that would be considered uncivil in one environment (e.g., on a news site’s comment section) might be seen as perfectly appropriate in another (e.g., in a private chat group). For example, uncivil discourse has been found to be equally common on Facebook and Twitter, but intolerant discussions occur less often on Facebook than on Twitter, where people communicate mainly with strangers16. Furthermore, what one person considers uncivil may be seen as civil by someone else1. Incivility is thus necessarily a subjective term, shaped not only by group norms, but also by individuals’ subjective evaluation of those norms.
In the current study, we use a similar definition of incivility as Frischlich et al.6. They define incivility as a communication that a person subjectively judges as having an unnecessarily disrespectful or offensive tone. The incivility can be aimed at other people, or towards specific topics and the content can be in any format, such as text, image, or video. With uncivil communication, we refer both to producing and spreading uncivil content6. Uncivil communication is set apart from a related term cyberbullying17 in that it is not necessarily intended to harm, and it is not necessarily a repeated behavior.
Online disinhibition
One of the reasons why uncivil online communication is so common is that online environments tend to lead to reduced behavioral constraints10,18. Simply put, people may do things online that they would not do in face-to-face interaction. This phenomenon has been named as the “online disinhibition effect”10. In the original description of the phenomenon, Suler10proposed that the online disinhibition effect leads to two, seemingly opposite behavioral outcomes. “Benign disinhibition” refers to seemingly positive outcomes, such as an increased tendency to share personal thoughts, or to help strangers in online settings. “Toxic disinhibition” refers to negative behavioral outcomes, such as an increased tendency to use rude language or to express hatred. These opposite outcomes, benign and toxic disinhibition are seen as two sides of the same coin, reflecting the same underlying behavioral disinhibition process10,19. Suler10proposed that online disinhibition is a result of several characteristics of online interactions that separate them from face-to-face communication (e.g., asynchronicity, relative anonymity, invisibility of the interaction partners). Later empirical research has confirmed that many of these characteristics indeed increase both uncivil behaviors (e.g18,20,21). , and experiences of online disinhibition22.
Importantly, features of the used online media are not the only factors determining the level of disinhibition, but disinhibition also varies substantially across individuals10,19. In other words, online environments lead to more behavioral disinhibition in some people than others. Individuals who report high levels of online disinhibition have an increased likelihood of expressing hate online23,24, and of engaging in cyberbullying25, trolling19, and online sexual harassment26. Experienced online disinhibition is typically measured with one of several questionnaires developed for this purpose (e.g19,25). , , as we also do in the current study.
Personality, empathy, and uncivil communication
Identifying factors that predict online disinhibition and uncivil behavior is crucial for understanding online incivility. There is ample evidence that individual’s personality and level of empathy play a role in influencing online behavior. Many of the studies that have examined the relationship between personality and online incivility have focused on the Big Five and Dark Triad models. It has been found that “dark personality traits” (i.e., psychopathy, narcissism, machiavellianism) correlate with uncivil online communication, aggressive online behavior, trolling, and experienced online disinhibition5–7,27. However, in the study of Koban et al.28 none of the Dark Triad Traits were statistically significantly related to the intention to comment uncivilly.
The five-factor model of personality29is one of the most comprehensive personality theories. In this model personality is described with five different dimensions: openness, conscientiousness, extraversion, agreeableness, and neuroticism. Prior research has shown that conscientiousness and agreeableness are negatively related, and extraversion is positively related to cyberbullying30,31. Additionally, Koban et al.28 found that openness and agreeableness negatively predicted uncivil commenting intentions.
Cognitive empathy refers to the ability to comprehend others’ affective states32–34. Affective empathy, on the other hand, means the capability to vicariously experience others’ affective states32,34,35. In the current study, we measured both of these core components: cognitive and affective empathy. It has been found that low cognitive empathy is associated with experienced online disinhibition in adolescents5. Additionally, Wright et al.23 found that callous-unemotional personality traits (characterized by lack of empathy, remorse, and guilt) predicted cyberbullying in adolescents.
Emotions, emotion regulation, and uncivil communication
While earlier research partially explains inter-individual variation in uncivil online behavior, much of this variation remains to be explained. We suggest that uncivil communication could further be explained by individual differences in emotion regulation processes. It is widely accepted that emotions are key motivators of intra- and inter-individual behavior, both in face-to-face interaction and on the internet36,37. In online discussions, emotional messages influence the direction the conversation will take. Emotionally toned messages generally cause emotional reactions in the readers38,39and may result in both further emotional content exchanges8,40and drift the discussion off-topic41. Further, if reading messages evokes intense negative emotions, people are more likely to speak out rather than remain silent regardless of the topic of the discussion or of how hostile or friendly the discussion climate is42. Experimental studies have also demonstrated a causal link between negative emotions and uncivil communication. Cheng et al.9 found that triggering negative emotions with experimental manipulations increased trolling when compared to triggering of positive emotions. Correspondingly, Seering et al.43 reported that user interface elements that evoke positive emotions as compared to no user interface intervention, increased the positivity of written messages. Taken together, these studies suggest that the more negative emotions people experience, the more likely they are to engage in uncivil communication.
Importantly, people are not passively at the mercy of their emotional reactions, but instead, they constantly regulate their emotions. Emotion regulation refers to the various psychological processes that modulate the quality, intensity, or duration of emotions44–46. This includes both explicit, effortful attempts at altering emotions, as well as implicit, relatively automatic emotion regulation processes46. Emotion regulation is important in practically all domains of life. Use of adaptive emotion regulation strategies, such as acceptance and cognitive reappraisal, predicts, for instance, better general well-being47, mental health48, and satisfaction of relationship49.
In one of the few papers that have investigated how emotion regulation processes shape online interactions, Fan et al.50investigated how the implicit emotion regulation process known as affect labeling (naming emotions) affects the emotions of Twitter users. They found that after users had named their emotional state, the tone of their subsequent posts changed towards a more neutral state than before naming of the emotion. This suggests that successful emotion regulation during social media use can lead users to act in a calmer manner. It could thus be expected that people with deficiencies in emotion regulation skills would be particularly prone to act uncivilly. Consistent with this, it was found that adolescents with emotion regulation difficulties are more likely than others to engage in cyberbullying51,52. However, to our knowledge, no studies have investigated associations between emotion regulation, online disinhibition, and uncivil communication.
The present study examined whether individual differences in emotion regulation beyond variation explained by personality and empathy would predict uncivil communication. We measured emotion regulation using two approaches. First and foremost, we measured difficulties in emotion regulation53,54. According to Gratz and Roemer54, difficulties in emotion regulation may occur in six domains: (1) nonacceptance of emotions, (2) difficulties in engaging in goal-directed behaviors when distressed, (3) difficulties in controlling impulses when distressed, (4) lack of awareness of emotions, (5) limited access to effective emotion regulation strategies, and (6) lack of clarity of emotions. In addition to emotion regulation difficulties, we also measured the habitual use of two different and most well-researched emotion regulation strategies: cognitive reappraisal and expressive suppression45,47. Reappraisal involves cognitive processing of emotion-evoking situations to alter their influence on emotions (e.g., interpreting that a hostile message was written by someone who was just having a bad day). Reappraisal is typically considered an adaptive emotion regulation strategy. Suppression, on the other hand, is characterized by concealment of emotional expressions (e.g., keeping a still facial expression when experiencing anger). Suppression is usually considered a relatively maladaptive strategy. There is growing evidence that reappraisal reliably downregulates negative emotional experiences, but suppression is not as effective55.
Methods
Overview of the study
We conducted an online survey to examine associations between emotion regulation, online disinhibition, and uncivil communication. Respondents reported whether they had engaged in uncivil communication during the past three months, and filled in scales measuring personality, empathy, experienced online disinhibition, difficulties in emotion regulation, and use of different emotion regulation strategies (reappraisal and suppression). As disinhibition is associated with incivility, we anticipated that online disinhibition could be a possible mediating mechanism by which emotion regulation would predict uncivil communication. Therefore, we subjected the data to a mediation analysis. We were particularly interested in whether emotion regulation would explain uncivil communication, beyond what is explained by personality and empathy. Therefore, the effects of personality and empathy were controlled for in the analysis.
Statement on research ethics
At the time of conducting this research, anonymous surveys did not require formal review by a research ethics committee under Finnish research governance, in which the Declaration of Helsinki defines applicability to research on identifiable human data. The survey followed ethical research practices as defined, for example, by the Finnish National Board on Research Integrity and Finnish law (i.e., voluntary participation; reassurance of anonymity, data protection, and confidentiality; advance information on purpose and content; provision of contact details of the research team; and full disclosure of involved organizations). This information was summarized on the first page of the online survey. Anonymous electronic consent to voluntary participate was required to begin the survey, but no signatures were obtained.
Sample
The sample size was determined based on a power analysis, conducted using the pwr package for R56. We calculated a sample size sufficient for detecting a typical correlation in an individual difference study (r= .1957;). We anticipated this to be a reasonably conservative estimate, as earlier studies on similar topics have reported larger correlations (e.g., r = .3 between emotion regulation difficulties and cyberbullying in Jiang et al.52). The power analysis suggested a sample size of N= 215 at α = 0.05 and power = 0.8058. This number of respondents were recruited via the participant recruitment platform Prolific59. The inclusion criteria were English as first language and a minimum of 90% approval rate on Prolific. Respondents received £1.88 as compensation for their participation. Eleven respondents failed at least in one of the three attention check items included in the survey60 and were thus removed from all analyses and replaced with new respondents. The median response time in the analyzed sample was 10 min and 42 s. The characteristics of the analyzed sample are presented in Table 1.
Table 1.
Characteristics of the sample.
| M | SD | |
|---|---|---|
| Age | 33.2 | 11.6 |
| Gender | n | % |
| Male | 78 | 36.3 |
| Female | 133 | 61.9 |
| Other | 4 | 1.9 |
| Education | n | % |
| Not applicable | 2 | 0.9 |
| Secondary | 19 | 8.8 |
| High school | 49 | 22.8 |
| Technical / community college | 23 | 10.7 |
| Undergraduate | 93 | 43.3 |
| Graduate | 26 | 12.1 |
| Doctorate | 3 | 1.4 |
| Frequency of posting content | n | % |
| Never | 61 | 28.4 |
| Less than once a month | 52 | 24.2 |
| At least once a month | 51 | 23.7 |
| At least once a week | 39 | 18.1 |
| At least once a day | 12 | 5.6 |
Survey
The respondents filled in a survey that was described as investigating “online and offline behavior”. The survey was conducted with the LimeSurvey web application. It consisted of scales measuring uncivil communication and various individual traits (see below). The survey also contained a few additional questions about whether they had read or posted user-generated content in the past three months, what types of platforms they had used, and whether they usually post to anonymous services or services that display users’ real names or nicknames. Only the scales on uncivil communication and individual traits are reported in the current study.
Outcome variable - uncivil communication
Uncivil communication was measured using the scale developed by Frischlich et al.6. In the scale, respondents evaluate the frequency of various kinds of responses to uncivil content, such as posting, “liking”, reporting, and ignoring such content during the past three months. The scale consisted of ten items, all on a 5-point scale (ranging from “Never” to “At least once per day”). We analyzed responses identically to Frischlich et al.6: In the analysis, we only included the four items representing uncivil communication, that is, how often the participants had posted, shared, showed, and liked uncivil content (Cronbach’s alpha = 0.672). While Frischlich et al.6 called this “uncivil participation,” we think that the term “uncivil communication” would be more appropriate as posting, sharing, and showing are generally active forms of communication, and liking, while leaning toward participation, can be a subtle form of communication. Just like in the earlier study, the distribution of the means of the items representing uncivil communication were highly positively skewed. Thus, just like Frischlich et al.6, we dummy-coded whether participants had engaged in these types of behaviors (0 = no, 1 = yes). In the current sample, 48.4% of respondents reported having engaged in uncivil communication in the past three months. In turn, we can see from Table 1 that 47.4% have posted at least once a month while 52.6% of the respondents have rarely (less than once a month) or never posted a comment online. Table 2 illustrates the connection between the frequency of posting and engaging in uncivil communication during the past three months. It shows that respondents having engaged in uncivil communication are not only people who post more often, but also many of infrequent commenters had engaged in uncivil communication.
Table 2.
Distribution of the respondents (%) regarding frequency of posting and engaging uncivil communication during past three months.
| Uncivil communication | |||
|---|---|---|---|
| No | Yes | ||
| Frequency of posting | Never | 41 | 20 |
| Less than once a month | 25 | 27 | |
| At least once a month | 24 | 27 | |
| At least once a week | 18 | 21 | |
| At least once a day | 3 | 9 | |
Mediating variable - online disinhibition
Online disinhibition was measured using the Measure of Online Disinhibition19. It consists of 12 items, including statements such as “I am more assertive online than I am offline” and “My behaviors online are less restricted than in person”, all evaluated on a 5-point Likert scale. All items load on the same factor (Cronbach’s alpha = 0.921). For the analysis, we averaged the items and then mean-centered the resulting online disinhibition score.
Predictor variables - emotion regulation
Difficulties in emotion regulation were measured using the brief version of the Difficulties in Emotion Regulation Scale (DERS-16)53. It consists of sixteen items (e.g., “I have difficulty making sense out of my feelings”, “When I’m upset, I have difficulty getting work done”), measured on a 5-point Likert scale. For the analysis, we averaged the sixteen items to create an index of difficulties in emotion regulation (Cronbach’s alpha = 0.945). The score was then mean-centered.
To measure the habitual use of different emotion regulation strategies, we used the Emotion Regulation Questionnaire (ERQ)47. The questionnaire consists of ten items measuring the use of cognitive reappraisal (Cronbach’s alpha = 0.881) and expressive suppression (Cronbach’s alpha = 0.723). The items representing the two factors were averaged to create the respective indices, and then mean-centered.
Covariates - personality and empathy
Personality traits were measured using the Big Five Inventory-1061. It is a very short, 10-item questionnaire measuring personality traits according to the five-factor model of personality. The short scale was chosen to keep the survey brief. All items are on a 5-point scale. Items were reverse scored where necessary and averaged to create indices of each of the five factors. The scores were mean centered for the analysis. Reliability of the factors were examined using the Spearman-Brown coefficient, as it is considered the best statistic for assessing the reliability of two-item scales62. Coefficients for the agreeableness and openness subscales were unacceptably low (0.384 and 0.350, respectively), and were thus excluded from the analysis. Coefficients for the other subscales were 0.612 for the extraversion, 0.523 for the conscientiousness, and 0.579 for the neuroticism. These coefficients are also quite low, but can be considered acceptable/sufficient63.
Empathy was measured with the Basic Empathy Scale in Adults32. The questionnaire consists of 20 items, measured on a 5-point scale. The items were reverse scored where necessary, and then averaged to create indices of affective empathy (Cronbach’s alpha = 0.866) and cognitive empathy (Cronbach’s alpha = 0.850). The scores were mean centered for the analysis.
Data analysis
We conducted a mediation analysis with a regression-based approach, using the PROCESS macro for SPSS (version 4.0)64. In this analysis, a mediation model is constructed by conducting a series of regression analyses to estimate associations between predictor variables, mediators, and outcome variables. Ordinary least squares and logistic regression models are used for continuous and dichotomous consequent variables, respectively. Covariates can be included in the mediation model to control their effects. Indirect effects are calculated by taking the product of the (predictor variable → mediator) and (mediator → outcome variable) regression coefficients. Confidence intervals (CI) for the indirect effects are estimated with bootstrapping.
We constructed a single mediation model with three predictor variables (emotion regulation difficulties, reappraisal, suppression), one mediator (online disinhibition), one outcome variable (uncivil communication, dummy-coded variable) and five covariates (conscientiousness, extraversion, neuroticism, cognitive empathy, affective empathy). We examined 95% CIs for the indirect effects based on 100,000 bootstrap samples.
Results
Tables 3 and 4 report the means and standard deviations of each studied variable and single correlation coefficients between variables as basic statistics.
Table 3.
Mean and standard deviation of the studied variables.
| Mean | Std. Deviation | |
|---|---|---|
| Extraversion | 2,93 | 0,95 |
| Agreeableness | 3,46 | 0,83 |
| Conscientiousness | 3,60 | 0,85 |
| Neuroticism | 3,22 | 1,01 |
| Openness | 3,57 | 0,88 |
| Disinhibition | 2,27 | 0,92 |
| AffectiveEmpathy | 3,51 | 0,68 |
| CognitiveEmpathy | 3,97 | 0,55 |
| Reappraisal | 4,71 | 1,03 |
| Suppression | 3,95 | 1,19 |
| Emotion Regulation Difficulties | 2,41 | 0,87 |
| Uncivil (dummy) | 0,48 | -- |
Table 4.
Correlations of the variables.
| Extraversion | Agreeableness | Conscientiousness | Neuroticism | Openness | Disinhibition | Affective Empathy | Cognitive Empathy | Reappraisal | Suppression | Emotion Regulation Difficulties | ||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Agreeableness | −0,039 | -- | ||||||||||
| Sig. | 0,569 | |||||||||||
| Conscientiousness | 0,092 | ,267** | -- | |||||||||
| Sig. | 0,181 | 0,000 | ||||||||||
| Neuroticism | -,303** | −0,065 | -,316** | -- | ||||||||
| Sig. | 0,000 | 0,340 | 0,000 | |||||||||
| Openness | 0,000 | −0,097 | −0,032 | 0,105 | -- | |||||||
| Sig. | 0,996 | 0,156 | 0,639 | 0,124 | ||||||||
| Disinhibition | -,249** | −0,088 | -,295** | ,326** | −0,022 | -- | ||||||
| Sig. | 0,000 | 0,200 | 0,000 | 0,000 | 0,745 | |||||||
| Affective Empathy | −0,012 | ,177** | 0,048 | ,241** | ,247** | 0,011 | -- | |||||
| Sig. | 0,864 | 0,009 | 0,487 | 0,000 | 0,000 | 0,873 | ||||||
| Cognitive Empathy | 0,091 | 0,066 | ,210** | -,159* | ,211** | −0,016 | ,445** | -- | ||||
| Sig. | 0,184 | 0,337 | 0,002 | 0,020 | 0,002 | 0,818 | 0,000 | |||||
| Reappraisal | 0,095 | 0,105 | ,182** | -,245** | 0,074 | −0,066 | ,168* | ,363** | -- | |||
| Sig. | 0,164 | 0,125 | 0,008 | 0,000 | 0,283 | 0,338 | 0,014 | 0,000 | ||||
| Suppression | -,232** | −0,063 | -,139* | 0,072 | −0,076 | ,225** | -,168* | -,144* | 0,059 | -- | ||
| Sig. | 0,001 | 0,359 | 0,042 | 0,291 | 0,264 | 0,001 | 0,013 | 0,035 | 0,386 | |||
| Emotion Regulation Difficulties | −0,133 | −0,111 | -,364** | ,584** | 0,048 | ,367** | ,146* | -,188** | -,283** | ,279** | -- | |
| Sig. | 0,052 | 0,104 | 0,000 | 0,000 | 0,485 | 0,000 | 0,032 | 0,006 | 0,000 | 0,000 | ||
| Uncivil (dummy) | 0,007 | −0,069 | −0,102 | 0,037 | −0,082 | ,190** | −0,008 | 0,026 | -,154* | 0,085 | 0,127 | |
| Sig. | 0,916 | 0,315 | 0,134 | 0,588 | 0,230 | 0,005 | 0,911 | 0,700 | 0,024 | 0,214 | 0,062 | |
The associations between the emotion regulation variables (predictors), online disinhibition (the mediator), and uncivil communication (the outcome variable) are depicted in Fig. 1. Estimates of the indirect effects are shown in Table 5. For the regression model with disinhibition as the outcome variable, the model summary is F 7.9201, R 0.4850, R2 0.2352, MSE 0.6699, df1 8.0, df2 206.0, p 0.0000. For the model with incivility as the outcome variable, the model summary is −2LL 280.1480, Model LL 17.6773, df 9.0, p 0.0391, McFadden 0.0594, CoxSnell 0.0789, Nagelkrk 0.1053. Associations between the covariates (personality and empathy variables) and online disinhibition and uncivil communication are shown in Table 6.
Fig. 1.

The resulted mediation model. The solid and dashed lines indicate significant and non-significant associations, respectively. The numbers represent regression coefficients. Where uncivil communication is the consequent variable, the coefficients are in log-odds metric. Covariates are omitted from the figure for the sake of clarity. **p < .01, *p < .05.
Table 5.
Bootstrap estimates of indirect effects of emotion regulation variables on uncivil communication via online disinhibition. The coefficients are in log-odds metric.
| Effect | SE | 95% CI | |
|---|---|---|---|
| Emotion regulation difficulties | 0.098 | 0.069 | [0.001, 0.267] |
| Reappraisal | 0.014 | 0.030 | [−0.036, 0.086] |
| Suppression | 0.029 | 0.030 | [−0.016, 0.101] |
Table 6.
Coefficients for the associations between personality and empathy variables and online disinhibition and uncivil communication. Where uncivil communication is the consequent variable, the coefficients and confidence intervals are in log-odds metric.
| Online disinhibition | |||||
|---|---|---|---|---|---|
| Coefficient | SE | t | p | 95% CI | |
| Extraversion | −0.15 | 0.06 | −2.36 | 0.019 | [−0.278, −0.025] |
| Conscientiousness | −0.19 | 0.07 | −2.57 | 0.011 | [−0.327, −0.043] |
| Neuroticism | 0.13 | 0.08 | 1.72 | 0.087 | [−0.019, 0.278] |
| Affective empathy | −0.14 | 0.10 | −1.43 | 0.154 | [−0.342, 0.054] |
| Cognitive empathy | 0.25 | 0.13 | 1.97 | 0.051 | [−0.001, 0.494] |
| Uncivil communication | |||||
| Coefficient | SE | Z | p | 95% CI | |
| Extraversion | 0.14 | 0.166 | 0.86 | 0.391 | [−0.183, 0.467] |
| Conscientiousness | −0.11 | 0.188 | −0.57 | 0.572 | [−0.474, 0.262] |
| Neuroticism | −0.16 | 0.194 | −0.54 | 0.399 | [−0.545, 0.217] |
| Affective empathy | 0.00 | 0.259 | 0.00 | > 0.999 | [−0.508, 0.508] |
| Cognitive empathy | 0.43 | 0.327 | 1.32 | 0.186 | [−0.208, 1.073] |
When the scales with low internal consistency, agreeableness and openness, are included in the model, the indirect effect of emotion regulation difficulties on uncivil communication via online disinhibition is slightly smaller, and notably, the confidence interval includes zero (effect = 0.096, SE = 0.07, 95% CI [−0.00, 0.27]).
Most importantly, the model showed that emotion regulation difficulties predicted higher levels of online disinhibition, which, in turn predicted uncivil communication. The 95% CI for this indirect effect was entirely above zero. Additionally, the results showed that the use of reappraisal predicted lower levels of uncivil communication directly. No other significant associations involving emotion regulation variables were found. As for the covariates, it was found that extraversion and conscientiousness predicted lower levels of online disinhibition. No other significant associations were found.
Discussion
Our analysis implies a clear statistical association: the more the respondents reported emotion regulation difficulties, the more they experienced online disinhibition, that is, weakened online behavioral restraints. Higher online disinhibition then predicted a higher likelihood of reporting uncivil online communication. This result expands on earlier knowledge on the role of user emotions in shaping online behavior. Prior research suggests that encountering emotion evoking content can lead people to post emotional messages8, to troll other users9, and to shift the discussion off-topic41. It is likely that emotion regulation has a central role in this process: individuals who successfully manage to regulate their emotions are better in inhibiting their behavior than those not so successful in regulation. Better regulation in turn associates to calm and civil communication (see also50). In contrast, individuals who have difficulties with emotion regulation are at an increased risk to lash out, resulting in uncivil communication.
Importantly, the results pertaining to emotion regulation strategies showed that especially higher use of reappraisal predicted lower levels of uncivil communication directly. Suppression as a strategy was not statistically significantly associated to uncivil communication. Reappraisal is a cognitive-linguistic strategy in which people change their emotions by reformulating the meaning of an emotion-eliciting event65. It is considered an adaptive emotion regulation strategy because its frequent use is associated with enhanced control of emotion, interpersonal functioning, and well-being47. The results support the previous findings that reappraisal is an effective strategy in downregulating negative emotional experiences55, and suggest that reappraisal skills have important role in regulating emotions and affecting interpersonal communication also in online environments.
Importantly, the observed associations between emotion regulation and uncivil communication were significant even when controlling for personality and empathy. Previous research has demonstrated that dark personality traits and low empathy are associated with uncivil online behavior5–7. Further, previous research has found that some of the Big five personality traits are related to cyberbullying30,31or uncivil communication intentions28. The current result suggests that emotion regulation processes predict inter-individual variation in uncivil communication, beyond the variation in personality and empathy.
The personality and empathy measurements also produced noteworthy findings. High conscientiousness and extraversion predicted low levels of online disinhibition. This consolidates prior research, which has found that dark personality traits such as psychopathy correlate with online disinhibition5,7. Dark personality traits and traits of the five-factor model are distinct, but partially overlapping concepts. For instance, conscientiousness negatively correlates with dark personality traits66, and therefore the current results are in line with earlier research. In the present study neither cognitive nor affective empathy predicted online disinhibition or uncivil communication. This contrasts with one earlier study, which reported that cognitive empathy negatively correlated with online disinhibition in adolescents5. One possible explanation as to why the result was not replicated is the different samples (adolescents vs. adults).
Limitations, future work, and practical implications
A limitation of the study was that two of the five personality scales, agreeableness and openness, had unacceptably low internal consistency in our data, and were thus excluded from the analyses. Their exclusion is noteworthy, as openness and agreeableness may have impact on online behavior. However, we believe the exclusion of these scales was justified. While we used a personality scale which has demonstrated sufficient psychometric properties in earlier research61, the low internal consistency suggests the two scales did not satisfactorily measure the underlying personality dimensions in our study.
Additionally, while the three retained subscales—Extraversion, Neuroticism, and Conscientiousness—were included in the mediation model, their internal consistency was also quite low. It has been noted that these subscales are best represented by their 2-item versions61, which influenced our decision to include them. However, the low internal consistency of these scales should be taken into account when interpreting the results and understanding the connection between personality traits and uncivil communication. Future studies should consider using the original Big Five Inventory, consisting of 44 items, as this longer version has demonstrated higher reliability61.
Another limitation of our study is our use of Prolific to recruit responders. Prolific has limitations such as sample bias, where participants tend to be younger and more educated, limiting generalizability. Also, the response quality may have been affected by survey fatigue or speeding up in completing the survey. In hindsight, we believe it was good that we also included individuals who never posted online comments or posted rarely in the sample. While this may be considered as a limitation, we argue that online behavior is more than just posting comments. In the present study, almost half of the infrequent commenters reported having engaged in uncivil communication in the past three months. Instead of posting they have likely shared, showed, or liked other people’s uncivil content and thereby engaged in uncivil behavior. Nevertheless, future studies could focus on only individuals who are frequent posters to dwell deeper on the question of posting frequency and emotion regulation.
Based on the current study, we cannot draw firm conclusions about causal relationships between the success of emotion regulation and online behavior. However, experimental studies have established a causal link between users’ emotions and uncivil commenting behavior8,9. Thus, it seems likely that much of the uncivil communication online can result from a failure to regulate emotions. To determine that the associations between emotion regulation, online disinhibition and uncivil communication are causal in nature, experimental or longitudinal studies would be needed.
The findings of the present study not only broaden our understanding of who engages in uncivil online communication, but also suggest possible avenues for mitigating this type of behavior. We suggest development of interventions for promoting users’ emotion regulation, thereby potentially mitigating uncivil communication. Similarly, Kiskola et al.67recently proposed user interface designs that might help users recognize and regulate their emotions. Future user interface designs could make use of the current finding that especially reappraisal predicted low likelihood of uncivil communication. For example, when encountering emotion evoking content on social media, the user could be nudged to reappraise the intentions of the person who had posted the content, possibly reducing the inclination to respond uncivilly. Further, as the present study mostly focused on forms of incivility that relate to speech-based norms, and only indirectly addressed violations of the inclusion-based norms11, future studies could focus more on the inclusion based norms.
Conclusion
The current study showed that individuals with emotion regulation difficulties are particularly prone to engage in uncivil communication online. It seems likely that a significant portion of uncivil communication results from users’ failure to regulate their emotions. Thus, to mitigate some of the negative outcomes of social media, it would be important to develop methods to support emotion regulation while engaging in online interactions on social media.
Author contributions
A.S., M.I., T.O., J.K., A.R., P.I., and V.S. did study design, analysis and interpretation. G.B. did analysis and interpretation. A.S. and M.I. were the lead authors responsible for writing the main manuscript text. All authors reviewed the manuscript and approved the submitted version.
Data availability
The data that support the findings of this study are openly available in Fairdata.fi at https://doi.org/10.23729/de438b0d-072f-4f2e-a8b5-204576ab76ef.
Declarations
Competing interests
The authors declare no competing interests.
Footnotes
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
- 1.Coe, K., Kenski, K. & Rains, S. A. Online and uncivil? Patterns and determinants of incivility in newspaper website comments. J. Commun.64, 658–679 (2014). [Google Scholar]
- 2.Chen, G. M. & Lu, S. Online political discourse: exploring differences in effects of Civil and Uncivil disagreement in News website comments. J. Broadcast. Electron. Media. 61, 108–125 (2017). [Google Scholar]
- 3.Anderson, A. A., Brossard, D., Scheufele, D. A., Xenos, M. A. & Ladwig, P. The nasty effect: online incivility and risk perceptions of Emerging technologies. J. Comput. Commun.19, 373–387 (2014). [Google Scholar]
- 4.Prochazka, F., Weber, P. & Schweiger, W. Effects of Civility and reasoning in user comments on Perceived Journalistic Quality. J. Stud.19, 62–78 (2018). [Google Scholar]
- 5.Antoniadou, N., Kokkinos, C. M. & Markos, A. Psychopathic traits and social anxiety in cyber-space: a context-dependent theoretical framework explaining online disinhibition. Comput. Hum. Behav.99, 228–234 (2019). [Google Scholar]
- 6.Frischlich, L., Schatto-Eckrodt, T., Boberg, S. & Wintterlin, F. Roots of incivility: how personality, Media Use, and Online experiences shape Uncivil Participation. Media Commun.9, 195–208 (2021). [Google Scholar]
- 7.Kurek, A., Jose, P. E. & Stuart, J. I did it for the LULZ’: how the dark personality predicts online disinhibition and aggressive online behavior in adolescence. Comput. Hum. Behav.98, 31–40 (2019). [Google Scholar]
- 8.Kramer, A. D. I., Guillory, J. E. & Hancock, J. T. Experimental evidence of massive-scale emotional contagion through social networks. Proc. Natl. Acad. Sci. U S A. 111, 8788–8790 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Cheng, J., Bernstein, M., Danescu-Niculescu-mizil, C. & Leskovec, J. Anyone can become a troll: Causes of trolling behavior in online discussions. in Proceedings of the ACM Conference on Computer Supported Cooperative Work, CSCW vol. 2017 1217–1230Association for Computing Machinery, (2017). [DOI] [PMC free article] [PubMed]
- 10.Suler, J. The Online Disinhibition Effect. Cyberpsychology Behav.7, 321–326 (2004). [DOI] [PubMed] [Google Scholar]
- 11.Muddiman, A. Personal and public levels of political incivility. Int. J. Commun.11, 21 (2017). [Google Scholar]
- 12.Park, J. S., Seering, J. & Bernstein, M. S. Measuring the Prevalence of Anti-Social Behavior in Online Communities. Proc. ACM Human-Computer Interact.6, (2022).
- 13.Akhtar, S. & Morrison, C. M. The prevalence and impact of online trolling of UK members of parliament. Comput. Hum. Behav.99, 322–327 (2019). [Google Scholar]
- 14.Papacharissi, Z. Democracy online: civility, politeness, and the democratic potential of online political discussion groups. New. Media Soc.6, 259–283 (2004). [Google Scholar]
- 15.Rossini, P. & Beyond Incivility Understanding patterns of Uncivil and intolerant discourse in Online Political Talk. Communic Res.10.1177/0093650220921314 (2020). [Google Scholar]
- 16.Oz, M. & Nurumov, B. Intolerant versus uncivil: examining types, directions and deliberative attributes of incivility on Facebook versus Twitter. First Monday27, (2022).
- 17.Slonje, R., Smith, P. K. & Frisén, A. The nature of cyberbullying, and strategies for prevention. Comput. Hum. Behav.29, 26–32 (2013). [Google Scholar]
- 18.Lapidot-Lefler, N. & Barak, A. Effects of anonymity, invisibility, and lack of eye-contact on toxic online disinhibition. Comput. Hum. Behav.28, 434–443 (2012). [Google Scholar]
- 19.Stuart, J. & Scott, R. The measure of online disinhibition (MOD): assessing perceptions of reductions in restraint in the online environment. Comput. Hum. Behav.114, 106534 (2021). [Google Scholar]
- 20.Lowry, P. B., Zhang, J., Wang, C. & Siponen, M. Why do adults engage in Cyberbullying on Social Media? An Integration of Online Disinhibition and Deindividuation effects with the Social structure and Social Learning Model. Inf. Syst. Res.27, 962–986 (2016). [Google Scholar]
- 21.Rösner, L., Winter, S. & Krämer, N. C. Dangerous minds? Effects of uncivil online comments on aggressive cognitions, emotions, and behavior. Comput. Hum. Behav.58, 461–470 (2016). [Google Scholar]
- 22.Wu, S., Lin, T. C. & Shih, J. F. Examining the antecedents of online disinhibition. Inf. Technol. People. 30, 189–209 (2017). [Google Scholar]
- 23.Wright, M. F., Harper, B. D. & Wachs, S. The associations between cyberbullying and callous-unemotional traits among adolescents: the moderating effect of online disinhibition. Pers. Individ Dif. 140, 41–45 (2019). [Google Scholar]
- 24.Wachs, S. & Wright, M. F. Associations between Bystanders and Perpetrators of Online Hate: The Moderating Role of Toxic Online Disinhibition. Int. J. Environ. Res. Public Heal.15, 2030 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Udris, R. Cyberbullying among high school students in Japan: Development and validation of the online disinhibition scale. Comput. Hum. Behav.41, 253–261 (2014). [Google Scholar]
- 26.Zhong, L. R., Kebbell, M. R. & Webster, J. L. An exploratory study of technology-facilitated sexual violence in online romantic interactions: can the internet’s toxic disinhibition exacerbate sexual aggression? Comput. Hum. Behav.108, 106314 (2020). [Google Scholar]
- 27.Buckels, E. E., Trapnell, P. D. & Paulhus, D. L. Trolls just want to have fun. Pers. Individ Dif. 67, 97–102 (2014). [Google Scholar]
- 28.Koban, K., Stein, J. P., Eckhardt, V. & Ohler, P. Quid pro quo in web 2.0. Connecting personality traits and Facebook usage intensity to uncivil commenting intentions in public online discussions. Comput. Hum. Behav.79, 9–18 (2018). [Google Scholar]
- 29.McRae, R. R. & Costa, P. T. The Five-Factor Theory of Personality. in Handbook of Personality: Theory and Research, 3rd Ed. 159–181 (Guilford Press, 2008).
- 30.Festl, R. & Quandt, T. Social relations and Cyberbullying: the influence of individual and structural attributes on victimization and perpetration via the internet. Hum. Commun. Res.39, 101–126 (2013). [Google Scholar]
- 31.van Geel, M., Goemans, A., Toprak, F. & Vedder, P. Which personality traits are related to traditional bullying and cyberbullying? A study with the big five, Dark Triad and sadism. Pers. Individ Dif. 106, 231–235 (2017). [Google Scholar]
- 32.Carré, A., Stefaniak, N., D’Ambrosio, F., Bensalah, L. & Besche-Richard, C. The basic empathy scale in adults (BES-A): factor structure of a revised form. Psychol. Assess.25, 679–691 (2013). [DOI] [PubMed] [Google Scholar]
- 33.Hogan, R. Development of an empathy scale. J. Consult Clin. Psychol.33, 307–316 (1969). [DOI] [PubMed] [Google Scholar]
- 34.Jolliffe, D. & Farrington, D. P. Development and validation of the Basic Empathy Scale. J. Adolesc.29, 589–611 (2006). [DOI] [PubMed] [Google Scholar]
- 35.Bryant, B. K. An index of Empathy for children and adolescents. Child. Dev.53, 413 (1982). [Google Scholar]
- 36.Bradley, M. M. & Lang, P. J. Emotion and motivation. Handb. Psychophysiol. 581–607. 10.1017/CBO9780511546396.025 (2009).
- 37.Derks, D., Fischer, A. H. & Bos, A. E. R. The role of emotion in computer-mediated communication: a review. Comput. Hum. Behav.24, 766–785 (2008). [Google Scholar]
- 38.Syrjämäki, A. H. et al. Emotionally toned online discussions evoke subjectively experienced emotional responses. J. Media Psychol. Theor. Methods Appl. (2022).
- 39.Gervais, B. T. Incivility online: affective and behavioral reactions to Uncivil Political posts in a web-based experiment. J. Inf. Technol. Polit. 12, 167–185 (2015). [Google Scholar]
- 40.Chmiel, A. et al. Collective emotions Online and their influence on Community Life. PLoS One. 6, e22207 (2011). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Topal, K., Koyuturk, M. & Ozsoyoglu, G. Emotion -and area-driven topic shift analysis in social media discussions. in Proceedings of the IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining, ASONAM 2016 510–518 (Institute of Electrical and Electronics Engineers Inc., 2016). 10.1109/ASONAM.2016.7752283 (2016).
- 42.Masullo Chen, G., Muddiman, A., Wilner, T., Pariser, E. & Stroud, N. J. We should not get rid of Incivility Online. Soc. Media Soc.5, (2019).
- 43.Seering, J. et al. Designing user interface elements to improve the quality and civility of discourse in online commenting behaviors. Conf. Hum. Factors Comput. Syst. Proc. 1-14. 10.1145/3290605.3300836 (2019).
- 44.Gross, J. J. The emerging field of emotion regulation: an integrative review. Rev. Gen. Psychol.2, 271–299 (1998). [Google Scholar]
- 45.Gross, J. J., Es, J. M. & Ross, J. G. Emotion regulation: past, Present, Future. Cogn. Emot.13, 551–573 (1999). [Google Scholar]
- 46.Gyurak, A., Gross, J. J. & Etkin, A. Explicit and implicit emotion regulation: a dual-process framework. Cogn. Emot.25, 400–412 (2011). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 47.Gross, J. J. & John, O. P. Individual differences in two emotion regulation processes: implications for Affect, relationships, and well-being. J. Pers. Soc. Psychol.85, 348–362 (2003). [DOI] [PubMed] [Google Scholar]
- 48.Hu, T. et al. Relation between Emotion Regulation and Mental Health: A Meta-Analysis Review. 114, 341–362. 10.2466/03.20.PR0.114k22w4 (2014). [DOI] [PubMed]
- 49.Vater, A. & Schröder-Abé, M. Explaining the Link between Personality and Relationship Satisfaction: Emotion Regulation and Interpersonal Behaviour in Conflict Discussions. 201–215. 10.1002/per.1993 (2015).
- 50.Fan, R. et al. The minute-scale dynamics of online emotions reveal the effects of affect labeling. Nat. Hum. Behav.3, 92–100 (2019). [DOI] [PubMed] [Google Scholar]
- 51.Arató, N., Zsidó, A. N., Lénárd, K. & Lábadi, B. Cybervictimization and cyberbullying: the role of Socio-Emotional skills. Front. Psychiatry. 11, 465652 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 52.Jiang, Q. et al. Difficulties in emotion regulation and cyberbullying among Chinese adolescents: a mediation model of loneliness and depression. NP1105-NP1124. 10.1177/088626052091751737 (2020). [DOI] [PubMed]
- 53.Bjureberg, J. et al. Development and validation of a brief version of the difficulties in emotion regulation scale: the DERS-16. J. Psychopathol. Behav. Assess.38, 284–296 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 54.Gratz, K. L. & Roemer, L. Multidimensional Assessment of emotion regulation and dysregulation: development, factor structure, and initial validation of the difficulties in emotion regulation scale. J. Psychopathol. Behav. Assess.26, 41–54 (2004). [Google Scholar]
- 55.Kalokerinos, E. K., Greenaway, K. H. & Denson, T. F. Reappraisal but not suppression downregulates the experience of positive and negative emotion. Emotion15, 271–275 (2015). [DOI] [PubMed] [Google Scholar]
- 56.Champely, S. Basic Functions for Power Analysis [R package pwr version 1.3-0]. 10.32614/CRAN.PACKAGE.PWR (2020).
- 57.Gignac, G. E. & Szodorai, E. T. Effect size guidelines for individual differences researchers. Pers. Individ Dif. 102, 74–78 (2016). [Google Scholar]
- 58.Cohen, J. Statistical power analysis. Curr. Dir. Psychol. Sci.1, 98–101 (1992). [Google Scholar]
- 59.Palan, S. & Schitter, C. Prolific.ac—A subject pool for online experiments. J. Behav. Exp. Financ. 17, 22–27 (2018). [Google Scholar]
- 60.Oppenheimer, D. M., Meyvis, T. & Davidenko, N. Instructional manipulation checks: detecting satisficing to increase statistical power. J. Exp. Soc. Psychol.45, 867–872 (2009). [Google Scholar]
- 61.Rammstedt, B. & John, O. P. Measuring personality in one minute or less: a 10-item short version of the big five inventory in English and German. J. Res. Pers.41, 203–212 (2007). [Google Scholar]
- 62.Eisinga, R., Grotenhuis, M., Te & Pelzer, B. The reliability of a two-item scale: Pearson, Cronbach, or Spearman-Brown? Int. J. Public. Health. 58, 637–642 (2013). [DOI] [PubMed] [Google Scholar]
- 63.Taber, K. S. The Use of Cronbach’s alpha when developing and Reporting Research Instruments in Science Education. Res. Sci. Educ.48, 1273–1296 (2018). [Google Scholar]
- 64.Hayes, A. F. Introduction to Mediation, Moderation, and Conditional Process Analysis, Third Edition : A Regression-Based Approach. (2022).
- 65.Goldin, P. R., McRae, K., Ramel, W. & Gross, J. J. The neural bases of emotion regulation: reappraisal and suppression of negative emotion. Biol. Psychiatry. 63, 577–586 (2008). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 66.Jonason, P. K., Kaufman, S. B., Webster, G. D. & Geher, G. What lies beneath the dark triad dirty dozen: varied relations with the big five. Individ Differ. Res.11, (2013).
- 67.Kiskola, J. et al. Applying critical voice in design of user interfaces for supporting self-reflection and emotion regulation in online news commenting. in CHI’21: Proceedings of the CHI Conference on Human Factors in Computing Systems (Association for Computing Machinery, 2021). 10.1145/3411764.3445783 (2021).
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
The data that support the findings of this study are openly available in Fairdata.fi at https://doi.org/10.23729/de438b0d-072f-4f2e-a8b5-204576ab76ef.
