Abstract
News organizations’ online commenting tools have been touted as a boon for the deliberative process, yet only to the extent that they are used by a diverse group of participants who are civil and who provide information that enriches dialogue. The researchers in this study analyzed the degree to which posts in newspaper forums originated from frequent contributors and the civil and informational characteristics of those contributions. A content analysis of comments (N = 2,237) within forums adjacent to opinion articles on websites of 15 U.S. newspaper dailies was conducted to assess their civility (uncivil character attacks and uncivil language) and informational attributes. Frequent contributors—people who post often on a forum in a short period—were less likely to be civil and informational.
Keywords
The open exchange of ideas among people with varying perspectives is a long-standing feature of democratic political thought (Freeman, 2000; Papacharissi, 2002). The Internet enables citizens to engage in discussions to share their viewpoints (Chae, 2009; Chung, 2008; Delli Carpini, Cook, & Jacobs, 2004; Himelboim, 2010; Jansen & Koop, 2005; Light & Rogers, 1999; Schneider, 1996; Singer, 2009), and research has found that people are attracted to source diversity in online forums (Stromer-Galley, 2002). Ideally, a diversity of participants on newspaper websites would share information with each other to get a better understanding of public affairs, yet the presence of intolerance toward conflicting perspectives has also been observed (e.g., Hill & Hughes, 2001; Koop & Jansen, 2009). News organizations and researchers have expressed concerns related to the impact of uncivil comments on source participation and the usefulness of the information (Nielsen, 2012).
Frequent contributors could have a profound influence on perceived credibility, opinion climate, and openness of discussions on online forums (Koop & Jansen, 2009; Singer & Ashman, 2009). Back-and-forth discussions are highly valued in the deliberative model, which is the ideal communication environment to promote rational discussion among citizens (Wheatley, 2003). Yet there are drawbacks when a forum turns into a “more or less closed private community rather than open public community of discussion” (Chae, 2009, p. 18). Similarly, Thompson (2008) maintained that a discussion “does not count as deliberation at all,” if one or just a few people dominate the discussions (p. 501). Schild and Oren (2005) speculated that frequent contributors on political listservs might have exerted an “oligarchy of opinion” that may have curtailed participation from others in their study.
To begin to address these concerns, we assessed the content contributions of frequent contributors and compared them to less frequent posters. Scholars have begun to identify the social roles of individuals within online communities. Some roles identified include answer people, questioners, trolls, conversationalists, spammers, and flame warriors (Himelboim, Gleave, & Smith, 2009; Turner, Smith, Fisher, & Welser, 2005; Welser, Gleave, Fisher, & Smith, 2007). It has yet to be fully examined how content contributions vary from frequent and infrequent forum posters. Frequent posters as a collective may influence the direction of discussions within forums. To understand whether behaviors of frequent and infrequent forum posters differ, newspaper comment forums were assessed through a quantitative content analysis. The investigation involved recording the extent to which a few comment participants populated the comment forums adjacent to letters to the editor and editorial articles (N = 2,237) on 15 U.S. newspaper websites. This study compared post types (informational, uncivil character attacks, and uncivil language) of frequent contributors (posts ≥ 7) with infrequent ones.
Literature Review
Public Sphere
The public sphere is a concept of normative, theoretical, and empirical consequence (Mutz, 2002; Splichal, 2006). Howard (2011) defined the public sphere as “increasingly a digitally mediated space . . . in which people discuss cultural values, compose solutions to shared problems, and implement collective projects” (p. 40). It has proven to be popular for researchers to engage questions surrounding user-generated content on news sites and its attendant impact on the polity (Mitchelstein & Boczkowski, 2009).
The public sphere is a sounding board for all affected members of society to discuss and influence solutions to problems (Habermas, 1989, 1998). It retains its emancipatory potential because of its thrust toward encouraging individuals to have an equitable opportunity to take part in deliberations (Sinekopova, 2006). Modern democracies make the claim that political power ultimately resides with citizens, making the rights of free expression of paramount importance. Forums for free expressions are necessary because without access to information and opinion, it is difficult to form authentic political views (Sparks, 1989).
But is it possible for a community to be both socially cohesive and diverse, without leadership? Karppinen (2007) critiqued the naïve pluralism of public sphere discourse, questioning whether overemphasis on pluralism may ultimately result in unhealthy dissonance because participation does not equal to deliberation. Such a thing may be possible only if the community is both relatively small and homogenous. As Sulkin and Simon (2001) summarized, “Deliberative situations must be carefully structured in order to achieve the collective benefits described by the theorists” (p. 812). Thus, the identification of the roles that naturally evolve within these mediated spaces can help researchers assess whether diversity and discussion can coexist.
Online Forums as Public Spheres
While face-to-face interactions remain important, political conversations and socialization increasingly take place within electronic networks (Howard, 2011). The Internet shows particular promise for public dialogue because researchers have found that users may be more willing to express themselves online than they are face-to-face (Ho & McLeod, 2008; McDevitt, Kiouis, & Wahl-Jorgensen, 2003; Moe, 2008). Yet, most people who participate in forums on non-issue-specific websites are strangers, which may mean that hostility is more likely to be present because they are less concerned with relationship maintenance, identity development, and goal outcomes (Eveland, Morey, & Hutchens, 2011; Haythornthwaite & Wellman, 1998; Stromer-Galley, 2007). It is also difficult to build community when traditional news media websites feature episodic content, in which stories are framed as examples of discrete, disconnected events and topics (Iyengar, 1994).
The press has historically provoked political conversations and opinion formation, because it is believed that a diversity of voices participating in deliberations is considered a necessary condition for a healthy democracy (Anderson, Dardenne, & Killenberg, 1996; Carpenter & Blom, 2012; Ciofalo & Traverso, 1994; Commission on Freedom of the Press, 1947; Howley, 2007; Hujanen & Pietikäinen, 2004; Kim, Wyatt, & Katz, 1999; Wahl-Jorgensen, 1999). Today, discussion forums on newspaper websites provide opportunities for partisan groups to learn and talk about arguments, viewpoints, and philosophies that they may not have encountered before (Shah, Cho, Eveland, & Kwak, 2005). As long as Internet forum users follow comment policies, they can post as much as they want and react to numerous people involved in the deliberations.
However, journalists have expressed concern regarding the incivility and dominance of a few (Nielsen, 2012; Singer & Ashman, 2009) and uncertainty about the role of news organizations in reader spaces (Robinson, 2010). Kovach and Rosenstiel (2007) noted that online news today is more like an open-mic conversation rather than the historical one-way directional form of communication between journalists and their publics. In addition, news organizations must grapple with this amplified audience voice. They can cue comment members of the behavioral expectations, or they can play no leadership role. Newsroom commenting policies reflect disagreements on whether an organization’s role is simply to facilitate conversation or to help build community (Singer, 2006).
Frequent Contributors
Increased access does not mean that discourse will accurately reflect the public as a whole. The likelihood that people read and participate may be hindered if representations within mediated spaces are reflective of a vocal few. Previous research has shown the number of people engaged in online discussions is relatively small (Chae, 2009; Chung, 2008; Himelboim, 2010; Jansen & Koop, 2005; Light & Rogers, 1999; Schneider, 1996; Singer, 2009; Zhou, Chan, & Peng, 2008). Delli Carpini et al. (2004) found only 4% of Americans engaged in discussions in online discussion groups and 24% emailed or instant messaged people about policy issues several times a month. Zhou et al. (2008) found 43% of discussion threads featured only two to five writers related to their examination of public affair forums on one news website. Jansen and Koop’s (2005) research found inequality in posts showing that four users posted the majority of the comments on one forum. In Stromer-Galley’s (2002) analysis of a sample of U.S. citizens, she found 21% said that they had engaged in online dialogue related to politics. Pew Internet and American Life Project (2002) research showed that approximately 20% of U.S. adults posted comments in forums. Chae’s (2009) analysis of the online newspaper public forums found only one third of forums have more than 10 users leaving messages on the board, while 41% of forums had no more than 5 users having posted their messages on the forum.
Frequent contributors often are not labeled in research because they are typically not the focus of a study, but they have been referred to as active participants, leaders, discussion catalysts, and heavy posters (Hill & Hughes, 2001; Himelboim et al., 2009; Jansen & Koop, 2005; Turner et al., 2005). For the purposes of this study, discussion participants are separated into groups only based on their frequency of commenting. People that comment often (posts ≥ 7) are referred to as frequent contributors. Research has yet to fully assess the behavior of frequent posters. The goal of this study is to measure the extent to which frequent contributors proportionately appear within the comment forums on newspaper websites and to analyze the types of posts they contribute in comparison to infrequent contributors.
Forum Perceptions
The perception of one’s environment can affect the extent to which one contributes information socially, according to the spiral of silence theory. The theory states that people constantly monitor their environment to understand how to behave socially. A challenge to one’s convictions can create self-inconsistency, which can lead to interpersonal conflict. To restore one’s self-image, research has shown that people will distance themselves from attitudinally dissimilar persons or react aggressively toward people who do not align with their moral beliefs (Noelle-Neumann, 1974, 1993; Skitka, Bauman, & Sargis, 2005). People with strong convictions are less likely to conform to group pressures (Hornsey, Majkut, Terry, & McKimmie, 2003). The willingness to speak out in the face of opposition is related to whether they perceive the other side as good or bad (Noelle-Neumann, 1974, 1993). A reduction in cues related to the identity of a source leads people to question the competence of the online anonymous contributor, which fuels the exchange of critical and stereotyped reactions (Connolly, Jessup, & Valacich, 1990). And thus, the investigation of social roles can help identify in which situations uncivil communication takes place. In this study, frequent contributors are examined to determine whether their behavior is truly different from other posters. Frequent contributors may influence the behavior of other contributors and potential participants. Research shows people feel intimidated to post within forums (Light & Rogers, 1999). Scholars have raised concerns that some participants are overshadowing other contributors to such an extent that many people are turned away from participating in discussions. Schultz (2000) illustrated this with a reaction from a disillusioned former user of an online comment section: “I have used the Forum less and less, mainly because there is a hard core of individuals who apparently have little else to do, and appear, at length, on almost every forum I have taken a look at” (p. 215).
Incivility
It may not only be the sheer frequency of other people’s interactions within discussions that turns people off; it could also be that the negative tone of those comments is considered inappropriate and counterproductive for deliberation of social issues. Hibbings and Theisse-Morse (2002) concluded that imperfect deliberation might be more problematic than no deliberation.
Some people who post comments in online spaces enjoy expressing their opinion and hearing the opinions of others in a constructive and practical way (Hill & Hughes, 2001; Light & Rogers, 1999; Stromer-Galley, 2002, 2007). Others—even when it is a small minority of all forum users—can exhibit uncivil reactions that try to humiliate, threaten, or annihilate other forum members, because they may be intolerant of criticism directed at their own opinions, which can lead to name calling and flaming. Flaming consists of verbal attacks intended to offend either people or organizations. They are often in the form of profanity or personal attacks (Ferber, Foltz, & Pugliese, 2006; Reinig & Megias, 2004). These communication acts could have negative consequences for the participation and perceptions of all forum users (Papacharissi, 2004; Rosenberg, 2004).
While online discourse can be filled with disagreements, research shows that such interactions have been found to be mostly civil. Characteristics of uncivil posts could include hostility, aggression, intimidation, insults, offensive language, uninhibited behavior, sarcasm, or an unfriendly tone. An uncivil post was defined as a post that featured a character criticism or offensive language. Koop and Jansen (2009) examined political forums and noticed unconstructive engagement. And Hill and Hughes (2001) found 18% of Usenet messages within ideologically consonant groups were flames and 32% of posts in ideologically opposite groups were flames, yet Papacharissi (2004) found discussion in political newsgroups was largely civil.
To date, researchers have not examined how frequent and infrequent contributors of online discussion spaces differ in their communication behavior. Such a study is warranted because of the possibility that frequent contributors are more uncivil in their reactions to other discussants. Based on the spiral of silence theory, it is speculated that more frequent contributors will post more uncivil posts because people who tend to comment tend to be more vocally defiant (Noelle-Neumann, 1974, 1993). To assess their behavior, two forms of incivility (uncivil character attacks and uncivil language) were further explored. First, an uncivil post was operationalized as attacking the character of participants, news media, or the writer of an article, rather than the content (viewpoints) of others. Second, uncivil language was assessed by measuring how often posts contained content such as profanity, racial slurs, or shouting.
Based on the literature review, it is expected that frequent contributors will use both tactics more than infrequent contributors. Therefore, it is hypothesized,
Hypothesis 1: Frequent contributors will post more uncivil attacks than infrequent contributors in comment forums adjacent to opinion articles on U.S. newspaper websites.
Hypothesis 2: Frequent contributors will post more uncivil language than infrequent contributors in comment forums adjacent to opinion articles on U.S. newspaper websites.
Information
It is also important to examine whether frequent contributors support the public sphere by sharing information in a positive manner. Such large involvement in public discussions could be beneficial when information is shared in a civil manner, but there has been more attention to potential negative consequences (e.g., Koop & Jansen, 2009; Light & Rogers, 1999). For deliberation, it is helpful when discussants provide information from trusted sources to bolster their claims or dispute claims by others. Informational posts can include personal experiences, news media, books, government sources, speeches, and so on (Carpenter, 2010; Davis, 1999; Stromer-Galley, 2007). For instance, users can add other news media reports or data from other sources by linking to content. Himelboim et al. (2009) discovered that sharing information was key to encouraging engagement among other posters, however the practice was rare. People who posted factual information received more replies in comparison to other posters. Therefore, the last hypothesis is as follows:
Hypothesis 3: Frequent contributors will post more information in comment forums adjacent to opinion articles on U.S. newspaper websites than infrequent contributors.
Method
Sampling Procedure
The data came from 15 U.S. daily newspapers during a constructed week during a month-long period (April 7 to May 4, 2009). The comment post was the unit of analysis. Each night, between 8:00 p.m. and 11:59 p.m. American Eastern Standard Time, two coders downloaded all opinion articles from the newspaper websites. The download order of the newspapers was randomly adjusted for each night of the constructed week.
A total of 80 newspapers from a list of 190 dailies from the Audit Bureau of Circulations were used. Newspaper outlets from the three largest metropolitan areas were removed (Chicago, Los Angeles, and New York) because those news media markets are atypical in audience size and competition level. Of the newspapers, 20 were excluded because they did not have comments during the examined time period, leaving 60 dailies.
To make the project more manageable, three more steps were taken to reduce the number of opinion articles and comments analyzed. First, the newspapers were divided into categories based on circulation size: (a) more than 170,000 (large), (b) 50,001 to 170,000 (medium), and (c) 50,000 or less (small; St. Cyr, Carpenter, & Lacy, 2010). Second, newspapers were selected that had both staff editorials and letters to the editor within the sampled period. Third, newspapers were chosen based on their geographical diversity. Each state was limited to one newspaper, and all four U.S. regions were represented: Northeast (3), Midwest (3), West (5), and South (4). This approach produced 15 U.S. online newspapers representing 15 different states (Norwich, CT; Des Moines, IA; Casper, WY; Lima, OH; Rochester, NY; Lakeland, FL; Boise City, ID; Vallejo, CA; Lebanon, PA; Detroit, MI; Spokane, WA; Greenville, NC; Nashville, TN; Boulder, CO; and Elizabethtown, KY).
This sampling procedure resulted in 2,237 comments that were adjacent to 215 letters to the editor and editorials. The number of comments per newspaper ranged from 12 to 652. The mean number of comments per article was 37.5, with a standard deviation of 40.7 comments. If a conversation was already in progress, the discussion participants were able to post a new comment to the discussion, as well as react directly to a previous comment. Consequently, the latest post would be added to the conversation, but would feature the initial comment first (in most cases in italics and/or different color) followed by the reaction. Thus, participants were able to communicate directly to each other within the large discussion, if they desired to.
Coding Categories
Frequent contributors
No operational definition exists to act as a guide on how to categorize and identify frequent contributors. For the purpose of this study, the top 10% of users with the most comments were considered frequent members for their higher participation in comparison to other forum contributors. The coders identified the names of individuals and determined which users posted seven or more times. For each comment, the user name was put in a database. There were no identical user names across newspapers. This created a group of 75 people. This unit was compared to the other forum users who commented fewer than seven times.
Uncivil attack posts
Posts were classified under this category if they featured attacks of the participants’ character, news media, or the writer of an article. Three different types of attacks were coded: expressing negative character attacks toward article opinion writer (e.g., “This editorial is written by a complete idiot.”), negative character attacks toward news media (e.g., “The Free Press does not know what it is talking about, as usual.”), and negative character attacks toward other contributors (e.g., “This comment shows you are dumb and clueless.”). A comment that contained at least one of those attacks was labeled as an uncivil attack post.
Uncivil language posts
This variable is concerned with the tone of the post, which could include profanity, racial slurs, and shouting. Comments were analyzed to see if they contained profanity, expletives, or letters followed by dashes (e.g., “I just don’t have time for this @&#$%! right now!”), words in all caps that were used to mimic shouting (e.g., “
Informational posts
Comments were coded as to whether information was present or absent. For example, forum users can cite documents (“One recent study showed that if you eat genetically modified. . . .”), refer to personal sources (“Contact this number to find . . . : XXX-XXX-XXXX.”), describe experiences from self or others (“Please be careful. My cousin tried to get obtain a marriage license in New York and was turned down.”), or provide clarification (“Lipitor is a drug that is used to lower cholesterol.”). Informational posts also could include hyperlinks.
Intercoder reliability
Two graduate student coders coded 9.5% (n = 213) of the comments to establish intercoder reliability. Scott’s pi computation was selected for the nominal-level variables. Intercoder reliability ranged from .86 to 1.0 for the variables mentioned above.
Analysis
A chi-square test of independence compared the rates of uncivil attack, uncivil language, and information posts for the frequent and infrequent commenter groups.
Results
The 2,237 analyzed comments were linked to 776 unique forum users, for an average of 2.9 posts per person (SD = 4.4). The median was one post per person. Frequent contributors (9.7% of all participants) were responsible for a large share of all posts (46.8%). The data also indicated that 47.8% of the last 10 comments of all articles were from those frequent contributors. This means that this group continued to be active at the end of comment discussions.
One of the most frequent users—one with the nickname angry on Tennessean website—posted 46 times in the constructed week. To put that in perspective, with that number of replies he or she would be on pace for 2,392 comments per year. Another 12 forums users, including Abatha (Des Moines Register), saltshaker (Detroit Free Press), and Ithinkonmyown (Tennessean), commented more than 20 times. In fact, together with angry, they were responsible for 15.2% of all responses, while representing a mere 1.6% of all contributors within the constructed week.
Hypothesis 1 posited that frequent contributors would post more uncivil character attacks than infrequent contributors in comment forums adjacent to opinion articles on U.S. newspaper websites. Frequent contributors provided more uncivil attacks (18.0%) than infrequent discussion participants (16.7%); however, there was not a statistically significant difference, χ2(1, N = 2,237) = 0.65. One item measuring uncivil character attacks showed that frequent contributors (13.9%) contained more uncivil attacks toward comment posters than infrequent contributors (9.2%). Thus, frequent contributors were significantly more likely to engage other comment posters in an uncivil manner.
Hypothesis 2 posited that frequent contributors would post more uncivil language than infrequent contributors in comment forums adjacent to opinion articles on U.S. newspaper websites. Frequent contributors did not post more uncivil language (17.0%) than infrequent contributors (17.5%) in comments, χ2(1, N = 2,237) = 0.13. Of items measuring uncivil language, the use of all caps to mimic shouting was the most common way to express incivility; frequent members (13.8%) used all caps only slightly more than did infrequent members (13.0%). All differences were not statistically significant.
Hypothesis 3 reflected the prediction that frequent contributors would post more informational comments than infrequent ones. The hypothesis was not supported. The data indicate that of the 1,047 comment posts by frequent contributors, they posted statistically significant fewer informational posts (13.2%) than of the 1,190 posts created by infrequent contributors (19.6%), χ2(1, N = 2,237) = 16.64, p < .001. Additional analysis showed only 1.9% of the comments contained an informational hyperlink to another site within the constructed week: 17 links were to traditional news media sites and 3 were to blogs.
Discussion
This research mirrors the desire to understand whether news forums reflect the ideals of the public sphere. Online discussions are considered healthy and functioning when they are civil and informational. The results indicate that an overwhelming majority of the participants avoid uncivil language and character attacks, but at the same time comments that provide supplementary information were relatively sparse. This finding is in line with the previous research in this area that found a relatively small percentage of uncivil posts (e.g., Hill & Hughes, 2001; Koop & Jansen, 2009; Papacharissi, 2004) and a small presence of information within forums (Zhou et al., 2008). It is important to note, however, that even if a forum contains a small percentage of informational-type posts, such posts can still be very valuable to all parties involved because they can educate other forum participants on the issue or refer to other sources with additional information about the topic. For instance in this research, Des Moines Register’s reader sanityindesmoines wrote, “Well said, Sir. I highly recommend a piece on Huffintonpost.com today by Drew Westen, Ph.D., Professor of Psychology and Psychiatry at Emory University” (April 17, 2009, 3:26 p.m.). This is a good example of how a poster provided additional information linking to an external source. Such information could allow for continuation of the discussion, which gives participants additional opportunities to understand different viewpoints on important public affairs topics—and also may help newspapers to secure more loyal readers who spend more time on their websites.
When frequent and infrequent posters were compared, there was no significant difference in the use of uncivil language. However, frequent posters were significantly less likely to share information within forums. Furthermore, frequent posters were more prone to engage their fellow commentators rather than the newspaper or the article writer specifically. This suggests that frequent posters may treat the online forum as a community in and of itself.
Newspapers have professional and financial reasons for attracting frequent posters because of the media organizations’ desire to increase page views. Yet the frequent posters did not foster an environment in which they welcome meaningful engagement from other people. Therefore, one of the implications of this research is that people who post often could possibly not be readers whom news organizations want to cultivate, because frequent contributors attack other users more often. Instead, newspapers may want to look for a way to make it more appealing for (current) infrequent posters to become more engaged in online discussions.
The challenge of news organizations is to identify people or experts who can nourish an environment that represents the public necessity of truth and understanding. Providing information requires more knowledge and effort on the part of participants. To encourage such contributions, posters could be rewarded through public acknowledgments awarded by a news organization that hosts a forum or by other posters through virtual badges (or other visual markers). As some previous research has shown, people may contribute information more often when they perceive that it enhances their social standing (e.g., McLure-Wasko & Faraj, 2005).
Conclusion
This study had some important limitations that should be considered in interpreting its results. For instance, the newspaper sample was relatively small compared to the overall circulation of newspapers in the United States, although an effort was made to include newspapers that varied in (print) circulation and location. The analysis covered only a short period, which may not have been representative of the overall picture of newspaper commenting over a longer period. Finally, the study’s focus on opinion articles leaves a question as to whether comments on news stories display different informational and civility characteristics.
While online comment forums were touted as carrying the potential of bolstering the public sphere as envisioned by Habermas and others, the results of this research raise questions as to the real-world potential of these forums. This study examined two types of users: frequent and infrequent contributors. Frequent contributors seem to treat the online forum as a space for social networking, but their contributions are less informational relatively in comparison to infrequent contributors. As a result, the frequent contributors may discourage participation by those who are less interested in online community-building aspects of the forum and more interested in discussing issues under examination.
It would be of interest for future researchers to survey frequent and infrequent contributors of news forums to assess the variations in their motivations. Based on our sample, it does not appear that news forum users engage often with one another. If the appeal of forums is socially rooted, news organizations need to identify the leaders who can assist in the creation and maintenance of an online community of readers and commentators.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
