The Internet: a failed public sphere? The disengaging and mistrusting effects of incivility in online political discussion
Master’s Thesis
Matthijs Johannes de Vries (11626828) University of Amsterdam
Graduate School of Communication Master’s programme Communication Science
Abstract
While the Internet was initially lauded as a possible public sphere where everyone could debate political issues equally, later criticism considered online discourse to be highly uncivil, damaging the medium’s deliberative potential. In a sample of US citizens, exposure to incivility was found to significantly lower both an individual’s trust in others to do what is right and trust in others to overcome their political differences. In addition, exposure to an uncivil comment lowered an individual’s willingness to engage in political discussion and increases the likelihood of the individual to stop engaging in political discussion. Finally, a marginal moderating effect of partisanship was found on the effects of incivility on the latter two variables. These effects confirm earlier literature about how incivility carries significant negative effects on the Internet’s potential to contribute to deliberative democracy and suggest a need for policies designed to counteract incivility.
The Internet: a failed public sphere? The disengaging and mistrusting effects of incivility in online political discussion
With the rise of the Internet as one of the major means of communication, there has been a great deal of scholarly debate about the effects of this rise and the potential that the medium has as a tool for deliberation and news exposure (Dalisay, Kushin & Yamamoto, 2017; Gervais, 2015; Neblo, 2010; Reynolds, 2006). Particularly in the early years of the internet, many scholars lauded the idea of the Internet as a way to create a pluralistic public sphere in the style of Habermas (1991) and Calhoun (1992), where all citizens could engage in political discussion equally, in a civil manner, with attention given to all viewpoints regardless of popularity or position of its proponent.
Although the general opinion remained positive throughout the early 2000s (see for instance Armstrong & Zúniga, 2006; Dahlgren, 2005; Reader, 2012; Reynolds, 2006; Wright & Street, 2007;), more critical studies pointed out that online communication tended to be highly uncivil, with political discussions being rife with insults, name-calling, even hate speech and calls for violence (Borah, 2012; Kurtz, 2012). This became more problematic as the Internet became an even more popular outlet for discussing politics, surpassing the traditional newspaper and radio (Pew Research Center, 2016)., In addition, rather than the Internet creating a pluralistic atmosphere, some studies found that online discourse has become even more partisan than discourse on other forms of media (Baum & Groeling, 2008). The consensus in current literature is that the unregulated nature of the Internet caused partisanship and incivility to take control over any kind of meaningful discourse (Cicchirillo, Hmielowski & Hutchens, 2015), which opens possibilities for abuse and damages the Internet’s deliberative potential.
Recent attention to abuse on the Internet has for the most part been focused on trying to prevent large scale misinformation campaigns from happening (see for instance Allcott &
Gentzkow, 2016; Mele et al, 2017). However, the measures proposed (such as the “fake news” label implemented by Facebook) may only solve the symptoms of what made this misinformation successful, rather than the root causes. These root causes may lie in some fundamental aspects of how contemporary social media work and how its users interact with the site and each other. There appears to be a tendency of Internet users to only seek out news and political opinion that they agree with, partly caused by algorithms designed to maximize engagement, but also because incivility has become an expected part of online discourse (Cicchirillo, Hmielowski & Hutchens, 2015), which makes people tend to avoid discussion with members of political out-groups (Soffer & Gordoni, 2018). This in turn creates and reinforces an increasingly polarized electorate, which is a breeding ground for ideologically motivated misinformation and creates an environment that is too hostile for deliberation to occur.
This study concerns itself with the effects of this hostile environment, and how the Internet’s predilection towards uncivil (political) discourse carries consequences that reach far beyond the online realm. As incivility is generally perceived to be an inevitability of online discussion (Suler, 2004; Coe, Kenski & Rains, 2014; Santana, 2014; Gervais, 2015), it follows that an increase in the use of the Internet as a means of communication will also lead to more incivility, leading to stronger effects from incivility on society. Whereas most studies on (online) incivility so far have focused on under what conditions incivility is most likely to occur (Suler, 2004; Coe, Kenski & Rains, 2016), how people respond emotionally to incivility (Suhay, Bello-Pardo & Maurer, 2017), as well as the effects it has on political figures (Teocharis et al., 2016), this study will build upon existing literature and investigate two effects that are largely overlooked by earlier studies: lowered trust and disengagement from political discussion. Several studies already found that incivility in online discussion can shape perceptions of the subject the incivility is about (Anderson et al., 2014), increases attention given to a comment but decrease its
deliberative potential (Kwon & Cho, 2015), and increases (perceived) polarization of the public (Suhay et al., 2017), but there has been little attention paid to the effects of incivility on the online discussion itself, and what the effects are on the likelihood of someone to participate in the discussion. Since inclusivity as well as the presence of a ‘domain of common concern’ are crucial for a deliberative public sphere as described by Habermas (1991), it is of vital importance that a public sphere should contain as many willing participants as possible and that these participants can trust others to act in good faith towards addressing a common concern. Willingness to participate in discussion as well as trust in other participants therefore lie at the heart of determining the deliberative potential of the Internet. The aim of this study is therefore to examine the effects that incivility can have on a person’s actions, and subsequent consequences for online discussion and the Internet’s deliberative potential.
Online Disinhibition
While the study of online incivility could go back as far as the 1990s (see for instance Kayany, 1998), there was rarely any major attention given to this topic until the early 2000s. The apparent tendency of internet users to show more uncivil behavior online was first described by Suler (2004), who named this the online disinhibition effect. Two seemingly opposing types of disinhibition present in online discussion are distinguished. The first is benign disinhibition, which entails the increased expression of personal feelings and attributes (p. 321). Internet users are more likely to share emotions, wishes and views online compared to real life. The second type of disinhibition is what Suler (2004) named toxic disinhibition, which includes rude language, harsh criticism, hate speech and threats, but also a likelihood of internet users to seek out content they would not seek out in the real world, such as violent and sexual content. The
distinction between both types can be ambiguous at times, but it is especially the latter type with which this study is concerned.
Disinhibition is therefore stronger if certain factors are met, and in the case of political discussion, toxic disinhibition is especially strong when participants are anonymous. According to Santana (2014), non-anonymous comments to online newspaper articles were three times as likely to be civil compared to anonymous comments on the same topic (p. 28). But there are other factors playing a part in the prevalence of incivility as well. First, the subject matter is relevant to studies on incivility. Incivility is more prevalent in political discussions compared to nonpolitical discussions, but the perceived acceptable level of incivility is much lower in the former than in the latter (Kwon & Cho, 2015). Within the realm of political discussion, the issue at hand matters as well. Weightier topics, topics with clear “sides” and topics with high-profile political figures (for example the President) generate the most incivility (Coe, Kenski & Rains, 2014). Finally, the medium with which the internet is accessed matters. In a study of Twitter comments, tweets that were sent from a mobile device were more likely to be uncivil compared to tweets sent from a personal computer (Groshek & Cutino, 2016). This suggests that the rise of mobile communication devices also contribute to online disinhibition. Disinhibition on mobile devices is caused by the same factors as disinhibition on non-mobile devices (for example anonymity, invisibility and distance) but strengthened by the greater accessibility of mobile networks compared to wired networks, as well as greater levels of impulsivity, individualization and personalization that comes with using mobile devices to participate in discussion (Groshek & Cutino, 2016).
The Effects of Incivility
Disinhibition and its main causes are why online communication tends to contain more incivility than other forms of communication, but this study is also concerned with the effects of online incivility. Effects of incivility in mass media have been studied for decades, particularly television (see Elving, 1994; Cappella & Jamieson, 1997; Patterson, 1993), but the effects of online incivility differ on several points from the effects of incivility on television. First, the internet provides a more interactive environment for media consumers. Whereas television can only have a passive audience that receives but cannot respond to (uncivil) media content, the internet allows for immediate response to content, be it through comment sections, blog posts or social media. This may lead to more personal investment into media content, which may lead to stronger effects of incivility. Second, incivility on television is usually restricted to incivility coming from public figures, such as politicians and pundits. In the case of the internet, anyone can have an audience to vitriol, which may mean that established effects of incivility on television, such as decreased trust in politics (Cappella & Jamieson, 1997; Mutz & Reeves, 2005) apply differently to the online realm, for instance that incivility coming from ordinary voters may lead to decreased trust in other people in general.
One of the main reasons why incivility became more prevalent on television, is that it increases the attention given to the content and promotes interest in the topic at hand (Mutz & Reeves, 2005). This effect is also relevant on the Internet, where the marketplace of ideas is huge, and every participant vies for attention. It is therefore no surprise that uncivil web content generally gets more attention from users than civil web content (Kwon & Cho, 2015). But there are also negative aspects of this increased attention. As stated before, incivility on television decreases political trust, but in the case of online incivility, the presence of uncivil content increases an individual’s risk perception towards a political issue that the uncivil content is about
(Anderson et al., 2014). For instance, exposure to an article about nanotechnology, followed by an uncivil comment about this issue, makes an individual more likely to consider the issue of emerging technology to be more dangerous than when they are exposed to a civil comment under the same article. Anderson et al. refer to this phenomenon as the “nasty effect”, in which an individual’s perceptions can be shaped by incivility, even when this incivility is presented under neutral and civil information.
In addition to the ‘nasty effect’, scholars have also identified three other negative effects of incivility in online discussion, with which this study is especially concerned. The first effect is polarization, mainly caused by the increased use of internet for gathering and discussing news and the resulting increased exposure to (uncivil) partisan content (Baum & Groeling, 2008). The scope and scale of the internet allows for every internet user to seek out the information that they wish to obtain and ignore all other information. In many cases, this is magnified by search algorithms that filter out results that an individual is less likely to click on, based on earlier search and browsing history. This leads to an increase in selective exposure, in which an individual is only exposed to information that agrees with their beliefs, is exposed less to information that does not agree with their beliefs and is also less exposed to other individuals that they do not agree with. This selective exposure was initially facilitated by the rise of political weblogs, which use more partisan filtering in deciding which content they place (Baum & Groeling, 2008). These blogs also contain significantly more incivility in the user comments, when compared to user comments on mainstream media outlets (Seely, 2017). The “nasty effect” as described by Anderson et al. (2014) also increases polarization along the lines of issue familiarity, issue support and personal traits like religiosity. The reason for this is that those who are unfamiliar with an issue beforehand will rely on heuristics to form judgments on that issue, and these three heuristics are particularly important when exposed to hostile communication. Online partisan
criticism in the form of user comments also leads to affective polarization, defined as an increasingly hostile attitude towards political “out-groups” by an individual. Affective polarization increases as the exposure of partisan criticism towards an “opposing” political figure increases partisan attitude in favor of an individual’s political “in-group” (Suhay et al., 2017). Furthermore, while according to Hwang, Kim & Huh (2014), exposure to uncivil online discussion does not lead to personal attitude polarization, it does affect the perceived polarization of the public and lower expectations about public deliberation. Building forward on this lowered expectation, this study hypothesizes that exposure from incivility leads to a lower trust in people to overcome their political differences.
H1: Exposure to uncivil replies compared to civil replies in an online discussion will lead to a decreased trust in citizens to overcome their political differences.
Another negative effect from online incivility is a decrease in trust in a more generalized sense. Trust can be operationalized as political trust (or efficacy), which is the degree of how strongly an individual trusts the government or politicians to do the right thing (Mutz & Reeves, 2005), but it can also be defined as how much an individual trusts other citizens. In the case of the former, there is some evidence that political incivility in general decreases political trust (Mutz & Reeves, 2005). This effect is extensively found in studies about incivility on television, but some studies also find circumstantial evidence for a decrease in trust caused by incivility in the online realm, especially when the incivility is aimed towards a politician that the individual does not support (Suhay, Pardo & Maurer, 2017). There is also some circumstantial evidence that online incivility decreases trust in citizens in general. For instance, according to Chen and Ng (2016), the persuasive strength of an online comment decreases when its content is considered
uncivil. This may indicate that people generally perceive others to be untrustworthy when their comments are uncivil. However, this contradicts the “nasty effect” by Anderson et al. (2014), which states that incivility can shape perceptions of an individual. So more research on the effects of incivility on trust is necessary. The second hypothesis will therefore be based partially on the definition of political efficacy used by Mutz and Reeves (2005) but will replace trust in politicians with trust in citizens.
H2: Exposure to uncivil replies compared to civil replies in an online discussion will lead to a decreased trust in citizens to do what is right.
Third, online incivility can cause some users to stop participating in discussion or not engage in discussion at all, damaging the internet’s deliberative potential. This deliberative potential is already hampered by a lower likelihood of responding to a discussion, when the majority of participants is perceived to disagree with the user, due to fear of isolation (Soffer & Gordoni, 2018). In addition, there is a tendency of conflict-avoidant people to participate in online discussion in the first place (Dalisay, Kushin & Yamamoto, 2017), and adding incivility to these discussions make these people more likely to stop engaging in online discussion even when they do participate. This in turn will leave only those users who are more prone to incivility engaging in online discussion, possibly compounding the issue over time, especially as incivility is increasingly seen as acceptable behavior online (Cicchirillo, Hiemelowski & Hutchens, 2014). Incivility already affects the use of social media by political figures: since they are prime targets of incivility, they increasingly use their accounts as broadcasting tools, minimizing engaging with voters as this makes them even bigger targets for uncivil responses (Teocharis et al., 2016). But there is also evidence of regular citizens disengaging from online discussion whenever that
discussion becomes uncivil. Neblo et al. (2010) find that willingness for deliberation is particularly high among those who do not typically participate in partisan politics, and that much of the partisan bickering found online disengages these people from joining the conversation. In addition, Gervais (2015) finds that disagreeable incivility increases dissatisfaction with a message board’s discourse, and Ziegele, Weber, Quiring and Breiner (2017) find that incivility is a detrimental discussion factor that makes users less willing to respond to a comment, and that other detrimental factors make a discussion more likely to be uncivil. This study therefore hypothesizes that exposure to an uncivil reply leads to a decrease in willingness to engage or continuing to engage in discussion. As unwillingness to participate in discussion and unwillingness to continue participating may carry different contexts, this effect will be divided in two hypotheses for this study.
H3: Exposure to uncivil replies compared to civil replies in an online discussion will lead to a decreased willingness to participate in that discussion.
H4: Exposure to uncivil replies compared to civil replies in an online discussion will lead to an increased willingness to stop participating in that discussion.
Finally, partisan identity may play a role in all these effects. Strong partisans may be more strongly affected by affective polarization as described by Suhay et al. (2017), as their belief system makes them less likely to identify with the “other” party. Second, strong partisans are more likely to consume partisan media such as blogs, which are more likely to contain incivility (Baum & Groeling, 2008). Third, strong partisans are more likely to see their beliefs as part of their identity, making them more susceptible to incivility towards out-groups (Miller & Conover, 2015) and tend to react emotionally to attacks on these beliefs, for instance by replying
with incivility. Although Rains, Kenski, Coe and Harwood (2017) find that partisan identity is only partially related to incivility, they also find that incivility in online discussion from strong partisans increases when more members from the “other” group are present, presumably out of a desire for intergroup distinctiveness (p. 173). Finally, exposure to like-minded incivility increases the use of uncivil behavior (Gervais, 2013). Though all these studies hint at the possibility of partisan identity moderating the effects of incivility on political discourse, there is very little concrete theoretical evidence found in the literature so far. This study would therefore also like to answer the question on whether strong partisans are actually more likely to respond to a discussion that contains incivility and less likely to stop responding to a discussion that has turned uncivil.
RQ: Does partisan identity moderate the effects of exposure to an uncivil comment on trust and willingness to participate in an online discussion?
Methodology Sampling
This study employed an experimental design in the form of an online survey. To test the hypotheses, a convenience sample of at least 200 United States citizens had to be drawn. The reason for this minimum sample size was to create a sample that was large enough to reliably test the hypotheses, but small enough to be achievable with the study’s limited resources. To achieve this sample size, the survey used for the study was first spread around among the researcher’s social networks, which yielded 58 full responses. After that, respondents were gathered from the online crowdsourcing service Amazon Mechanical Turk, bringing the total amount of responses to 471. From these responses, a total of 379 responses were fully completed.
Although there are several drawbacks from the use of a convenience sample, such as a lack of representativeness, the study employed this method for two reasons. First, the study itself was conducted by a non-United States citizen living outside of the United States, making it nearly impossible to create a representative sample. Second, a convenience sample needs very little resources to gather respondents, making this method suitable for studies with little means that require smaller samples.
The main reason for exclusively selecting United States citizens was to measure the mediating effects of partisan identity more reliably. The United States has a very strong two-party system, allowing for a simple method of creating “in-groups” and “out-groups” that (Miller & Conover, 2015). In addition to the two-party system, the choice for Unites States citizens as sample was also motivated by the fact that incivility is a relatively well-established part of politics in the United States (Cappella & Jamieson, 1997), and that the internet hosts a large part of political discourse in the US in general: in 2016, 38% of United States citizens used the Internet to gather online political information (Pew Research Center, 2016). The study did not measure the difference in effects between Republican and Democratic voters, first because the sample was disproportionately made up of Democrats compared to Republicans, opening the possibility for skewed results, second because a neutral stimulus allows for a more general conclusion of the effects of incivility and the moderating effect of partisanship.
Measurement: Incivility
The survey is designed to be as short as possible, to maximize the response rate to this survey and prevent participants from abandoning the survey early. To test the effects of incivility, there of course needs to be one experimental group within the survey that is exposed to incivility, but there is also a need for a clear definition of incivility before a manipulation can be created.
Several scholars have debated about this definition (see for instance Papacharissi, 2004), but for the purpose of this study, incivility is defined as vitriol, rude critiques and name-calling, as well as the use of swearing and slurs. This is close to the definition used by Jamieson (1997), and excludes “flaming”, which is included by Papacharissi (2004) and defined as making outrageous claims and incensed discussion. The reason for not including this into the definition is because the mechanics behind “flaming” differ significantly, for instance in that this type of incivility often does not directly address another participant and may therefore also have significantly different effects on the discussion and its participants. Insults towards a political group is perceived differently by individuals than mockery of a political group, for instance (Wang & Silva, 2014). For this reason, incivility in this study will be defined as the use of personal attacks and insults aimed towards another person in an online discussion.
Demographics
A total of 379 responses from the survey were used for the final data analysis. Of these respondents, 352 (92.9%) were registered to vote, and most were registered in California (11.6%), Texas (10,6%), Florida (6.3%), New York (5.5%) and Pennsylvania (4.2%). 54.9% of the respondents were women, 44.1% were men. In terms of party identification, 41.4% identified as Democrats, 23.7% identified as Republican, 30.1% were independent and 4.7% identified with another party. All respondents were between 18 and 80 years old, with an average age of 37 years.
Finally, the survey also asked if the respondents used social media to discuss politics. This could be on any website, such as Facebook, but also comment sections or forums. 16.9% of respondent indicated they discuss politics on social media daily, 13.2% said they did this 4-6
times a week, 15.3% indicated they discuss politics 2-3 times a week, 27.2% only once a week, and 27.4% never discuss politics on social media.
Survey Design
To test the four hypotheses, this study will conduct several independent sample t-tests with exposure to the (un)civil comment as grouping variable. Participants were randomly assigned to one of two groups within the survey: one was exposed to a screenshot of an online discussion that contained incivility, and one was exposed to a screenshot of an online discussion that did not contain incivility. This discussion consisted of one “parent comment”, which was an argument about the topic of gun control, and one “child comment”, which is a reply to that argument. It was the latter “child comment” that was manipulated to be either civil or uncivil. The choice for the topic of gun control for this study was motivated by the fact that this topic is a common political subject. This makes the issue salient for many respondents, easily recognizable, and requiring little explanation, which increases its validity. For the same reason, the topic on gun control was one of two issues used in the study by Wang & Silva (2018). After being exposed to either one of the comments, the survey will ask several questions designed to test the hypotheses. These are all single-item measures, as using only a single measure makes the survey shorter, it requires less resources and is more accessible and less monotonous to respondents. The first question asks about the respondent’s feelings towards the comment, requiring them to fill in if the comment made them happy, sad, angry, surprised, afraid or disgusted on a scale of 1 through 10. This is not used to answer the research question, but serves as survey filler, to prevent respondents from immediately deducing the study’s aim and provide socially desirable answers. The second question asks the respondent whether such a comment made them likely to join that discussion, which was answered on a Likert scale from one (Strongly agree) through 5
(Strongly disagree) and is used for the third hypothesis (M = 2.62, SD = 1.33). The third question asks whether such a comment makes the respondent likely to stop responding to a discussion after receiving such a reply, answered on the same 5-point Likert scale (M = 2.78, SD = 1.35). As has been stated before, this study distinguishes between willingness to participate in discussion and likelihood to stop engaging in discussion, as there are different psychological mechanisms at work between the two. This distinction is also strengthened by the measurements used, as a reliability analysis reveals that there is little reliability between the two measurements (Spearman-Brown coefficient = 0.294).
The third and fourth questions measured political trust, respectively asking the respondent whether the comment they were exposed to decreased their trust in other citizens to do what is right (M = 2.76, SD = 1.20) and whether it decreased their trust to overcome their political differences (M = 2.35, SD = 1.17). Both were answered with a 5-point Likert scale. The final question is the manipulation check, which asked the respondent whether the comment they were exposed to was uncivil, again on a 5-point Likert scale, ranking from 1 (Strongly agree) to 5 (Strongly disagree). The mean score on this variable was 2.78 with a standard deviation of 1.49. To determine whether or not the manipulation check was successful, a t-test was carried out. Those who were exposed to an uncivil comment were more likely to consider the comment to be uncivil (M = 1.87, SD = 1.11) compared to those who were exposed to the civil comment (M = 3.72, SD = 1.24) with a significance of .03, indicating that the manipulation check was successful.
The study will use a linear regression to analyze if there is a significant moderation effect of partisanship on the relationship between exposure to incivility and the four dependent variables that are used in the study. In order to operationalize partisan identity, two questions regarding party identification were added to the survey. One asked which party the respondent
identified with: the Republican Party, the Democratic Party, another party, or as an independent. The other asked how strongly they identified with that party, on a 7-point scale ranging from 1 (very strong) to 7 (very weak). In addition to the earlier filters used for the main hypotheses, those who did not identify as either Democratic or Republican were excluded from the sample that measured the moderation effect, bringing the sample size of this section to 167 respondents. This sample was then divided into a partisan and non-partisan group: those that answered that they identified “strongly” or “very strongly” with their respective party were added to the partisan group, all others to the non-partisan group. The reason for this division was to create two groups of an equal a size as possible: the partisan group contained 86 respondents (47 exposed to the uncivil comment, 39 to the civil comment) and the non-partisan group contained 84 respondents (57 exposed to the uncivil comment, 27 to the civil comment).
Case
The comments used in the study were based on and designed to look like two comments on the social medium Reddit, which was chosen for three reasons. The first reason is that Reddit is likely to follow the mechanics of online discussion forums more than other social media, as Reddit accounts are fully anonymous (unlike Facebook) and allow for longer, more public discussion (unlike Twitter). Second, the way comments on Reddit are laid out make it immediately obvious whether a comment is a reply to another comment, at least more so than traditional online discussion forums, Facebook or Twitter. The final reason is that Reddit is less known among the general population compared to Facebook or Twitter, but nonetheless one of the largest social media outlets in existence. Meanwhile, there have been relatively few studies on Reddit compared to Facebook and Twitter, necessitating more research on this specific medium. The wording of the discussion in both screenshots were identical to each other except for the
civil/uncivil manipulation. Both discussions had the same “parent comment”, both “child comments” showed disagreement to the parent comment, and in both cases the (anonymous) usernames, comment length and comment structure were similar. This was to avoid confounding variables and to ensure that any difference between the two groups was caused by the manipulation of the comment’s civility.
Results Hypotheses
The study’s first hypothesis states that exposure to uncivil replies in an online discussion will lead to a decreased trust in citizens to overcome their political differences. The results find that this is the case: those that were exposed to an uncivil comment were significantly less likely to trust other citizens to overcome their political differences (M = 1.92, SD = 0.90) than those who were exposed to a civil comment (M = 2.78, SD = 1.25; see Table 1). The first hypothesis is therefore confirmed.
Table 1: t-test results of trust to overcome political differences with exposure to (in)civility as independent variable
Uncivil Comment Civil Comment
Variable N M SD N M SD df t p
Trust 192 1.92 .90 187 2.78 1.25 338 7.638 <.001 p < .05
The second hypothesis concerned itself with the question of how incivility will affect trust in other citizens in a more general sense, which was operationalized as “trust to do what is right”. The results for this variable also yielded significant differences between the group exposed to
incivility and the group exposed to a civil comment. Those who were exposed to an uncivil comment were overall less to trust citizens to do what is right (M = 2.46, SD = 1.16) than those who were exposed to a civil comment (M = 3.06, SD = 1.16; Table 2). The second hypothesis is therefore also confirmed.
Table 2: t-test results of trust in other citizens to do what is right with exposure to (in)civility as independent variable
Uncivil Comment Civil Comment
Variable N M SD N M SD df t p
Trust 192 2.46 1.16 187 3.06 1.16 377 5.031 <.001 p < .05
After confirming the negative effects of incivility on trust, the third and fourth hypotheses were focused on what effects incivility has on an individual’s participation in online discussion. The third hypothesis stated that exposure to uncivil replies in an online discussion will lead to a decreased willingness to participate in that discussion. Again, the test finds significant results, with those exposed to the uncivil comment being less willing to participate (M = 2.30, SD = 1.28) compared to those exposed to the civil comment (M = 2.95, SD = 1.31).
Table 3: t-test results of willingness to participate in discussion with exposure to (in)civility as independent variable
Uncivil Comment Civil Comment
Variable N M SD N M SD df t p
Willingness to participate
192 2.30 1.28 187 2.95 1.31 377 4.888 <.001
The fourth and final main hypothesis predicted that exposure to uncivil replies in an online discussion will lead to an increased willingness to stop participating in that discussion. Here we also find a significant difference between the uncivil group (M = 2.31, SD = 1.26) and the civil group (M = 3.25, SD = 1.26). The full results of this test can be found in Table 4.
Table 4: t-test results of likelihood to stop participating in discussion with exposure to (in)civility as independent variable
Uncivil Comment Civil Comment
Variable N M SD N M SD df t p Willingness to stop participating 192 2.31 1.26 187 3.25 1.26 377 7.241 <.001 p < .05 Moderation
The first four tests all show highly significant effects of exposure to uncivil messages on trust in other citizens to do what is right, trust to overcome their political differences, willingness to participate and likelihood of an individual to stop participating after receiving such a message. However, further tests do not show a moderation effect of partisanship on the effects of incivility on the two trust variables.
To test the moderation effect, a multiple linear regression was calculated to predict trust in others to overcome their political differences, based on exposure to incivility, partisanship and the interaction between exposure to incivility and partisanship. A significant regression equation was found (F(3, 243) = 14.146, p <.000) with an R² of .149. It was found that trust in others to overcome their political differences significantly decreased in the group exposed to incivility compared to the group exposed to civility (β = -.308, p = .027). However, for the moderation
effect only the interaction between civility and partisanship is relevant, and the interaction between exposure to incivility and partisanship is not found to be a significant predictor of decreased trust (p = .563). No significant moderation effect is therefore found (Table 5).
Table 5: Results from linear regression analysis showing the moderation effect of partisanship on the relationship between exposure to (in)civility and trust in other citizens to overcome their political differences
B SE β p F R² Adj. R² ― 14.146 .149 .138 Constant 2.515 .227 .000 Civility -.689 .310 -.308 .027 Partisanship .094 .080 .104 .242 Civility * Partisanship -.063 .108 -.089 .563 p <.05
For the moderation effect of partisanship on the effects of exposure to incivility on trust in others to do what is right, another multiple linear regression was carried out. Here as well, a significant regression equation was found (F (3, 243) = 7.015, p <.001) and trust in others to overcome their political differences was found to significantly decrease in the group exposed to incivility compared to the group exposed to civility (β = -.370, p = .011). However, here as well the interaction between exposure to incivility and partisanship was not found to be a significant predictor (Table 6).
Table 6: Results from linear regression analysis showing the moderation effect of partisanship on the relationship between exposure to (in)civility and trust in other citizens to do what is right
B SE β p F R² Adj. R² ― 7.015 .080 .068 Constant 2.980 .248 .000 Civility -.868 .338 -.370 .011 Partisanship .025 .087 .026 .777 Civility * Partisanship .091 .118 .124 .439 p <.05
The third hypothesis of this study states that exposure to uncivil replies in an online discussion will lead to a decreased willingness to participate in that discussion, which was confirmed by the t-test, but this effect may be moderated by partisanship as well. For instance, the effect may be lessened or even inverted for those who strongly identify with a political party. Like the third hypothesis, the data for this test has been reversed in order to align with other tests. For this third linear regression, a regression equation was found as well (F(3, 243) = 11.178, p <.000), with an R² of .121. Willingness of a respondent to participate in the discussion decreased by .243 for the interaction effect between civility and partisanship. Like the previous analyses, the interaction between exposure to incivility and partisanship was not found to be a significant predictor under the normal significance level (p <.05), but unlike the previous regression analyses, there is a marginal significance found (p = .059), meaning that there may still be a moderation effect of partisanship on willingness to participate in discussion, albeit at a lower certainty of p < .1 (Table 7).
Table 7: Results from linear regression analysis showing the moderation effect of partisanship on the relationship between exposure to (in)civility and willingness to participate in discussion
B SE β p F R² Adj. R² ― 11.178 .121 .110 Constant 3.240 .270 .000 Civility -.094 .368 -.036 .799 Partisanship -.050 .095 -.047 .603 Civility * Partisanship -.243 .128 -1.894 .059* p <.05, *p <.1
Figure 1: Interaction table of linear regression analysis showing the moderation effect of partisanship on the relationship between exposure to (in)civility and willingness to participate in discussion
Results from this third regression therefore show only a marginal significance, indicating weak evidence for a moderation effect of partisanship on willingness to participate in discussion. The same is the case for the moderation effect of partisanship on likelihood to stop engaging in discussion. In this final regression analysis, a regression equation was found (F (3, 243) = 17.645) with an R² of .116. Here too, a marginal significance is found in the interaction between
1 1,5 2 2,5 3 3,5 4 4,5 5
Low Incivility High Incivility
W il li n gn ess to part icip ate Low Partisanship High Partisanship
exposure to civility and partisanship (p <.1), indicating weak evidence for a moderation effect (Table 8).
Table 8: Results from linear regression analysis showing the moderation effect of partisanship on the relationship between exposure to (in)civility and likelihood to stop participating in discussion
B SE β p F R² Adj. R² ― 17.645 .116 .105 Constant 2.882 .275 .000 Civility -.245 .376 -.092 .515 Partisanship .127 .097 .118 .191 Civility * Partisanship -.237 .131 .285 .071* p <.05, *p <.1
Figure 2: Interaction table of linear regression analysis showing the moderation effect of partisanship on the relationship between exposure to (in)civility and likelihood to stop participating in discussion
1 1,5 2 2,5 3 3,5 4 4,5 5
Low Incivility High Incivility
L ik eli h ood to stop p ar tic ip atin g Low Partisanship High Partisanship
Discussion
The aim this study was to provide insight into the damaging effects of incivility on the Internet’s deliberative potential, which was once considered promising but has since significantly weakened. There was attention given specifically to willingness of an individual to participate in online discussion as well as trust in other people, both considered to be important requirements for a public sphere in which deliberation can be facilitated (Habermas, 1991; Calhoun, 1992).
The four hypotheses, which stated that exposure to incivility will negatively affect several aspects of the Internet’s deliberative potential, were all confirmed to be true. These results are in agreement with more recent literature about online incivility (Coe, Kenski & Rains, 2014; Dalisay, Kushin & Yamamoto, 2017; Gervais, 2015; Santana, 2014). Being presented with an uncivil comment made a respondent less likely to trust people to overcome their political differences, less likely to trust people to do what is right, less willing to participate in that discussion, and more willing to stop participating in that discussion if they would ever receive that reply, compared to respondents who were exposed to a civil comment. These results are in accordance with the findings of Mutz and Reeves (2005) on the effects of political incivility on television and may be explained by a tendency of people to avoid conflict when possible (Dalisay, Kushin & Yamamoto, 2017), as well as lessened satisfaction with the medium when discourse is uncivil (Gervais, 2015). While the question on whether these effects were moderated by a strong partisan identity showed less significant effects, a marginal significance of the interaction effects between civility and partisanship could be found in the cases of willingness to participate and likelihood to stop participating in an online discussion.
While the negative effects of incivility were all found to be significant, the question on whether there is a moderating effect of partisanship was more difficult to answer. There was no
significant moderation effect of partisanship found on the effect of exposure to (in)civility, both on trust in others to overcome their political differences as well as trust in others to do what is right. However, on the two moderating variables of partisanship, on exposure to (in)civility on willingness to participate and on exposure to (in)civility on willingness to stop participating, there was only a marginal significance found. This implies that there is very weak evidence for a moderating effect of partisanship on these variables. That there is a marginal moderating effect found of partisanship is likely due to the fact that strong partisans are more likely to see their political beliefs as a part of their identity (Miller & Conover, 2015) and therefore more likely to disengage from a discussion if that identity is attacked.
While this study has yielded very significant results, the way the study has been design carries drawbacks that cannot be ignored. The first issue concerns sampling. Due to a lack of resources, this study was forced to use a convenience sample followed by online crowdsourcing, making the sample nonrepresentative to the general population. Studies that attempt to replicate this study’s findings should therefore use probability sampling for their research designs. A second sampling issue is that Democrats were overrepresented, making up over 40% of the sample. This is likely due to the initial convenience sampling, which skewed towards younger and more liberal individuals. Although this study was only concerned with strength of partisan identity and not differences between parties, effects on Democrats and Republicans could differ on particularly salient issues such as abortion and gun control (Rains et al., 2017). Further studies should therefore take specific partisan identity into account, both during sampling and methodology.
Another issue concerning the methodology is the artificiality of the environment the respondents were exposed to. The stimulus was a screenshot of a comment embedded within a survey, which creates an environment that is dissimilar to an online discussion platform, meaning
that respondents could be affected differently by the stimulus than if they were exposed to the comment in an environment that was more similar to genuine online discussion platforms. The findings of this study should therefore be replicated through a more interactive medium or a simulation of online discussion platforms. Third, due to the artificial environment of a survey, this study was unable to distinguish exposure to incivility aimed at the respondent from exposure to incivility aimed at someone else. Though the survey did ask the respondent to imagine receiving such a reply in the survey question on likelihood to stop participating, this method could possibly have led to satisficing answers and required respondents to artificially place themselves into a situation. A follow-up study that simulates an online discussion platform should therefore make the distinction between the respondents being merely exposed to incivility and being personally replied to with incivility. Finally, this study used single-item measures, making them more vulnerable to different interpretations of the item and random measurement errors. As single-item measures allow for a shorter survey that uses less resources and are less monotonous for respondents (Hoeppner et al., 2011) it was a justifiable choice for this study, but future studies should account for this vulnerability by using multiple-item measures.
The final issue concerning methodology is the conceptualization of incivility, which in this study is restricted to personal insults in order to gather clear effect results. As demonstrated by Papacharissi (2004), incivility online entails more than personal insults, such as the use of general but inflammatory language and incensed discussion (“flaming”). Further studies should investigate the effects of this broader type of incivility and whether its effects are similar.
A more positive observation from the study is that there was relatively low attrition among respondents, and that few people failed the manipulation check. This means that both the stimulus as the questions used are likely to be very reliable, and that the survey was accessible and easily answerable. The salience of the issue used in the stimulus (a discussion on the topic of
gun control) has likely helped the reliability and accessibility as well, as this was a particularly well-known topic that required little explanation to respondents.
The results of this study open an avenue for debate on how to improve the Internet’s deliberative potential, most importantly through attempts at eliminating incivility in online comment sections. One method of improvement could be the use of moderators to remove incivility from discussion, or by implementing an approval-based comment system that requires user comments to be submitted for review by staff before publication. There have been few studies on how the use of these methods mitigate the negative effects of incivility so far, but it has been found that the mere presence of comment moderation significantly reduces the perception of bias in science news (Yeo et al., 2017), indicating that moderation significantly alleviates the negative consequences of the ‘nasty effect’ as described by Anderson et al (2014). The effects of comment moderation and the implementation of Trust and Safety policies are therefore an interesting avenue for further studies on the effects of incivility and are likely to become even more important for online media outlets as they take a more prominent role in political discourse.
In conclusion, the presence of incivility and its effects is likely to be a major contribution for why the initially positive outlook of the Internet as a public sphere have not come to fruition. With the Internet becoming a more important medium than the newspaper or radio for politics, this also carries societal implications. For this reason, measures intended to counteract incivility online must be considered by media outlets and media users alike.
References
Allcott, H., & Gentzkow, M. (2017). Social media and fake news in the 2016 election. Journal of Economic Perspectives, 31(2), 211-36. https://doi.org/10.1257/jep.31.2.211
Anderson, A. A., Brossard, D., Scheufele, D. A., Xenos, M. A., & Ladwig, P. (2014). The “nasty effect:” Online incivility and risk perceptions of emerging technologies. Journal of Computer‐ Mediated Communication, 19(3), 373-387. https://doi.org/10.1111/jcc4.12009
Armstrong, J., & Zúniga, M. M. (2006). Crashing the gate: Netroots, grassroots, and the rise of people-powered politics. White River Junction, VT: Chelsea Green Publishing.
Baum, M. A., & Groeling, T. (2008). New media and the polarization of American political discourse. Political Communication, 25(4), 345-365.
https://doi.org/10.1080/10584600802426965
Berg, J. (2016). The impact of anonymity and issue controversiality on the quality of online discussion. Journal of Information Technology & Politics, 13(1), 37-51. https://doi.org/10.1080/19331681.2015.1131654
Borah, P. (2012). Does it matter where you read the news story? Interaction of incivility and news frames in the political blogosphere. Communication Research, 41(6), 809-827. https://doi.org/10.1177/0093650212449353
Calhoun, C. J. (1992). Habermas and the public sphere. Cambridge: MIT Press.
Cappella, J. N., & Jamieson, K. H. (1997). Spiral of cynicism: The press and the public good. New York: Oxford University Press.
Cicchirillo, V., Hmielowski, J., & Hutchens, M. (2015). The mainstreaming of verbally aggressive online political behaviors. Cyberpsychology, Behavior, And Social Networking, 18(5), 253-259. https://doi.org/10.1089/cyber.2014.0355
Coe, K., Kenski, K., & Rains, S. A. (2014). Online and uncivil? Patterns and determinants of incivility in newspaper website comments. Journal of Communication, 64(4), 658-679. https://doi.org/10.1111/jcom.12104
Chen, G. M., & Ng, Y. M. M. (2016). Third-person perception of online comments: Civil ones persuade you more than me. Computers in Human Behavior, 55, 736-742.
Dahlgren, P. (2005). The Internet, public spheres, and political communication. Political Communication, 22(2), 147–62. https://doi.org/10.1016/j.chb.2015.10.014
Dalisay, F., Kushin, M. J., & Yamamoto, M. (2017). The Demobilizing Potential of Conflict for Web and Mobile Political Participation. In Y. Ibrahim (Ed.), Politics, Protest, and Empowerment in Digital Spaces (pp. 52-71). Hershey, PA: IGI Global.
Elving, R. D. (1994). Brighter lights, wider windows: Presenting Congress in the 1990s. Washington D.C: American Enterprise Institute and The Brookings Institution.
Gervais, B. T. (2015). Incivility online: Affective and behavioral reactions to uncivil political posts in a web-based experiment. Journal of Information Technology & Politics, 12(2), 167-185. https://doi.org/10.1080/19331681.2014.997416
Groshek, J., & Cutino, C. (2016). Meaner on mobile: Incivility and impoliteness in communicating contentious politics on sociotechnical networks. Social Media + Society, 2(4), 1-10. https://doi.org/10.1177/2056305116677137
Habermas, J. (1991). The structural transformation of the public sphere: An inquiry into a category of bourgeois society. Cambridge, MA: MIT Press.
Hoeppner, B. B., Kelly, J. F., Urbanoski, K. A., & Slaymaker, V. (2011). Comparative utility of a single-item versus multiple-item measure of self-efficacy in predicting relapse among young adults. Journal of Substance Abuse Treatment, 41(3), 305-312. https://doi.org/10.1016/j.jsat.2011.04.005
Hwang, H., Kim, Y., & Huh, C. U. (2014). Seeing is believing: Effects of uncivil online debate on political polarization and expectations of deliberation. Journal of Broadcasting & Electronic Media, 58(4), 621-633. https://doi.org/10.1080/08838151.2014.966365
Jamieson, K. H. (1997). Civility in the House of Representatives: A background report. (APPC Report #10). Philadelphia: The Annenberg Public Policy Center at the University of Pennsylvania. Jenkins, M., & Dragojevic, M. (2013). Explaining the process of resistance to persuasion: A
politeness theory-based approach. Communication Research, 40(4), 559-590. https://doi.org/10.1177/0093650211420136
Kayany, J. M. (1998). Contexts of uninhibited online behavior: Flaming in social newsgroups on usenet. Journal of the American Society for Information Science, 49, 1135–1141.
Kwon, K. H., & Cho, D. (2017). Swearing effects on citizen-to-citizen commenting online: A large-scale exploration of political versus nonpolitical online news sites. Social Science Computer Review, 35(1), 84-102. https://doi.org/10.1177/0894439315602664
Kurtz, J. B. (2012). Civility, American style. Relevant Rhetoric, 3, 1-23.
Mele, N., Lazer, D., Baum, M., Grinberg, N., Friedland, L., Joseph, K., ... & Mattsson, C. (2017). Combating fake news: An agenda for research and action. Conference report. Cambridge, MA: Harvard Shorenstein Center.
Miller, P. R., & Conover, P. J. (2015). Red and blue states of mind: Partisan hostility and voting in the United States. Political Research Quarterly, 68(2), 225-239. https://doi.org/10.1177/1065912915577208
Mutz, D. C., & Reeves, B. (2005). The new videomalaise: Effects of televised incivility on political trust. American Political Science Review, 99(1), 1-15. https://doi.org/10.1017/S0003055405051452
Neblo, M. A., Esterling, K. M., Kennedy, R. P., Lazer, D. M., & Sokhey, A. E. (2010). Who wants to deliberate — and why? American Political Science Review, 104(3), 566-583. https://doi.org/10.1017/S0003055410000298
Papacharissi, Z. (2004). Democracy online: Civility, politeness, and the democratic potential of online political discussion groups. New media & society, 6(2), 259-283.
https://doi.org/10.1177/1461444804041444
Patterson, T. E. (1993). Out of order: How the decline of the political parties and the growing power of the news media undermine the American way of electing presidents. New York: Alfred Knopf.
Pew Research Center (2016). The modern news consumer: News attitudes and practices in the digital era. Retrieved from http://www.journalism.org/2016/07/07/pathways-to-news/
Rains, S. A., Kenski, K., Coe, K., & Harwood, J. (2017). Incivility and Political Identity on the Internet: Intergroup Factors as Predictors of Incivility in Discussions of News Online. Journal of Computer-Mediated Communication, 22(4), 163-178. https://doi.org/10.1111/jcc4.12191 Reader, B. (2012). Free press vs. free speech? The rhetoric of “civility” in regard to anonymous
online comments. Journalism & Mass Communication Quarterly, 89(3), 495-513. https://doi.org/10.1177/1077699012447923
Reynolds, G. (2006). An army of Davids: How markets and technology empower ordinary people to beat big media, big government, and other Goliaths. Nashville, TN: Thomas Nelson.
Rowe, I. (2015). Civility 2.0: A comparative analysis of incivility in online political discussion. Information, Communication & Society, 18(2), 121-138.
Santana, A. D. (2014). Virtuous or vitriolic: The effect of anonymity on civility in online newspaper reader comment boards. Journalism Practice, 8(1), 18-33. https://doi.org/10.1080/17512786.2013.813194
Seely, N. (2017). Virtual Vitriol: A Comparative Analysis of Incivility Within Political News Discussion Forums. Electronic News, 12(1), 42-61.
https://doi.org/10.1177/1931243117739060
Soffer, O., & Gordoni, G. (2018). Opinion expression via user comments on news websites: analysis through the perspective of the spiral of silence. Information, Communication & Society, 21(3), 388-403. https://doi.org/10.1080/1369118X.2017.1281991
Suhay, E., Bello-Pardo, E., & Maurer, B. (2018). The Polarizing Effects of Online Partisan Criticism: Evidence from Two Experiments. The International Journal of Press/Politics, 23(1), 95-115. https://doi.org/10.1177/1940161217740697
Suler, J. (2004). The online disinhibition effect. Cyberpsychology & Behavior, 7(3), 321-326. https://doi.org/10.1089/1094931041291295
Theocharis, Y., Barberá, P., Fazekas, Z., Popa, S. A., & Parnet, O. (2016). A bad workman blames his tweets: the consequences of citizens' uncivil Twitter use when interacting with party candidates. Journal of Communication, 66(6), 1007-1031. https://doi.org/10.1111/jcom.12259
Wang, M. Y., & Silva, D. E. (2018). A slap or a jab: An experiment on viewing uncivil political discussions on Facebook. Computers in Human Behavior, 81, 73-83. https://doi.org/10.1016/j.chb.2017.11.041
Wright, S., & Street, J. (2007). Democracy, deliberation and design: the case of online discussion forums. New Media & Society, 9(5), 849-869. https://doi.org/10.1177/1461444807081230
Yeo, S. K., Su, L. Y. F., Scheufele, D. A., Brossard, D., Xenos, M. A., & Corley, E. A. (2017). The effect of comment moderation on perceived bias in science news. Information, Communication & Society, 1-18. https://doi.org/10.1080/1369118X.2017.1356861
Ziegele, M., Weber, M., Quiring, O., & Breiner, T. (2017). The dynamics of online news discussions: effects of news articles and reader comments on users’ involvement, willingness to participate, and the civility of their contributions. Information, Communication & Society, 1-17. https://doi.org/10.1080/1369118X.2017.1324505
Appendix A: Screenshot of Civil Stimulus
