Michael Xenos is an assistant professor in the Department of Communication Arts at the University of Wisconsin-Madison. His research explores the effects of new media forms on civic engagement and democratic citizenship.
Address: Department of Communication Arts, University of Wisconsin-Madison, 6136 Vilas Communication Hall, 821 University Avenue, Madison, WI 53706
Errata: Erratum Volume 13, Issue 4, 1019, Article first published online: July 2008
This study explores the implications of political weblogs for theories of mediated public deliberation. Guided by contemporary questions surrounding the internet and the public sphere, we examine blog and newspaper coverage of the nomination and confirmation to the U.S. Supreme Court of Samuel Alito with an eye toward further development of theories of mass deliberation. Specifically, we pursue questions concerning volume of coverage, ideological polarization, and interactive features in the blogosphere, using newspaper coverage as a point of reference. Data come from content analyses of newspaper stories mentioning Alito in the headline or lead paragraphs from the initial nomination announcement through final confirmation, as well as archival impressions of blog posts featuring hyperlinks to the newspaper stories. Our analysis suggests that blogs may enhance as well as complicate processes of mediated deliberation. We conclude by discussing empirical and conceptual implications of these findings for future research on the role of blogs in the contemporary public sphere.
New Mediated Deliberation: Press and Blog Coverage of the Alito Nomination
Proponents of deliberative democracy have perpetually struggled with the problem of scale. The harsh reality is that while processes of deliberation may be manageable in small settings like a New England-style town meeting, deliberative discussion as traditionally conceived grows infinitely more difficult as the number of participants grows (Gastil 1993). As a result, scholarship on public deliberation has focused either on structured sessions in relatively small settings (e.g. Fishkin 1995; Gastil 2000), or among members of elite institutions of government, such as the U.S. Congress (Bessette 1994; Mucciaroni & Quirk 2006).
In a political system where mediated deliberation plays an important role, the rationality of collective political discussion and opinion is dependent on the quality of deliberative discussion among professional communicators in the mass media (Page & Shapiro 1992). Largely through case studies, Page and others have documented examples of successes within the overall system of mass mediated deliberation (Graber 2004; Page 1996; Simon & Xenos 2000, Zaller 2003). However, this research has also highlighted distinct problems with traditional news media as a primary source for political discussion among citizens. Specifically, Page noted the potential for media outlets to create a self-serving or “constructed debate” (in which coverage selectively highlights issues and perspectives that place the media outlet’s own editorial position at the center of a carefully balanced discussion), as well as broader concerns over whether the range of perspectives offered in mediated discussion is fully representative of the range of views among the public (1996). Further, Bennett et al. (2004) have demonstrated the potential for traditional news-gathering norms to be at odds with principles of deliberation.
In this article, we attempt to revisit questions concerning mediated deliberation in light of the significant developments in new media that have taken place since previous case studies. As a recent review of the literature points out, once dominated by mass media, the public sphere must now be thought of as a space in which interactive new media are also prominent (Dahlgren 2005). Written at the dawn of the internet age, early concept explications and analyses of mediated deliberation provide a unique point of departure for evaluating the ways in which fundamental shifts in the political communication environment may have contributed to significant changes in the public sphere, possibly alleviating some of the problems of reliance on mass media outlets to stimulate discussion. In particular we explore the potential contributions made by political weblogs (or blogs) to the greater system of mediated deliberation, conceiving of political bloggers as a unique new class of “professional communicators” (Page 1996). To do so, we conducted a case study of traditional press and blog coverage of a distinct political media event, the nomination and confirmation process that ultimately resulted in Samuel Alito’s ascension to the U.S. Supreme Court. Drawing on concepts of mediated deliberation along with the broader literature on the political effects of new media, we focused our inquiry on the differences between discussions of the Alito nomination and confirmation processes in old and new media, as well as the ways in which the two may work together in the contemporary public sphere. Through the analysis of this particular case, we hope to illustrate the possibilities for new patterns and dynamics within the broader system of mediated deliberation.
We focus on political blogs as a key element of the evolving system of mediated deliberation for three reasons. First, such outlets often provide direct commentary on both news events as well as news content. As more and more traditional news outlets offer online content, it is also not uncommon for political bloggers to include hyperlinks to text and increasingly video content in the course of discussing political events, as well as the very coverage of those events. Although survey and interview data are relatively scarce, exploratory studies by McKenna (2007) and others have documented the sense in which bloggers use their posts to actively participate in public discourse, often providing “rebuttals to assertions made by journalists and other public figures” and featuring excerpts and links to their objects of critique (McKenna & Pole 2004, p. 4). In this way, political blogs stand at the nexus of both ends of the mediated deliberation process. Simultaneously, bloggers may actively synthesize and represent focal points of broader political discussions as professional communicators (or in most cases, semi-professional or amateur communicators), while also directly engaging in discussion of points highlighted by more traditional political content providers of the public sphere in the mass media.
A second reason is that many blogs include comment features that enable users to easily join the discussion after only a few clicks of the mouse. In their content analysis of a random sample of blogs in 2003, Herring et al. (2004) documented the presence of comment features on 43% of blogs included in their study. In the case of political blogs, this feature adds yet another dimension to processes of mediated deliberation by enabling readers to potentially engage in the publicly mediated process in real time. With this new media form the basic infrastructure assumed by scholars such as Page (1996), which often features very little contribution from ordinary citizens, has thus potentially undergone significant change, raising a number of questions concerning the implications these changes might have for the concept of mediated deliberation.
A final consideration is that the user base and impact of political blogs has grown significantly in recent years, which further highlights their role as an important element of mediated deliberation processes. While it is still true that most Americans do not regularly read blogs, and as of late 2004 the Pew Internet and American Life project estimated that a majority of internet users still did not understand the term (Rainie, 2005), recent usage statistics suggest significant growth in blog activity. For example, by 2006, Pew surveys indicated that 39 percent of internet users read blogs, and the figure for keeping blogs had reached eight percent (Lenhart & Fox, 2006). Moreover, research suggests that those going online for political information, often from political blogs, are more likely to be “opinion leaders” who are critical in the processes by which discussion points raised in the media filter out to smaller discussions among average citizens (Shah & Scheufele, 2006). Additionally, research suggests that blog readers find the content they encounter on blogs to be more credible than information from other sources, including traditional media sources (Johnson & Kaye, 2004). For these reasons, we believe blogs constitute a significant change in the overall system of mediated deliberation and we seek to identify the nature of this change by examining how the blogosphere and the traditional press react to contemporary political events, as well as patterns in the discussion of traditional press coverage within the blogosphere.
Our findings in the example pursued here suggest that blogs may indeed offer something significantly new to the landscape of the public sphere, enhancing as well as complicating processes of mediated deliberation. To demonstrate this, we begin with a brief discussion of our rationale for selecting the Alito nomination and confirmation as a case for study. With this case in mind, we then consider how blogging could potentially affect the structure, content, and interactive nature of mediated political discussion. To assess these expectations, we employed a unique research design, which is explained prior to the presentation of our principal findings. We conclude with a discussion of the implications of these findings for future research on new media and processes of mediated deliberation.
The Nomination Battle Over Samuel Alito
In the late summer and early fall of 2005, a buzz of activity surrounded the composition of the Supreme Court. With Sandra Day O’Connor’s announcement, on July 1st, of her intent to retire from the court, as well as the death of Chief Justice William Rhenquist on September 3rd, there were ultimately three nomination battles that took place leading to the court’s current composition. At the outset, scholars of political communication in new media eagerly awaited the unfolding of these events. Recall that the last major Supreme Court nomination battle, which included media frenzy over Anita Hill’s allegations of sexual harassment by the nominee, Clarence Thomas, took place well before the internet as most users know it existed.
As scholars have observed, the process by which judicial appointments are considered has become more political than ever, now resembling an election more than a simple appointment process (Davis 2005). What was once a relatively quiet affair has now become a national media spectacle, complete with vigorous campaigns waged by political parties and interest groups to compete for public opinion and even the final outcome itself (Davis 2005; Parry-Giles 2006). Given the growth in prominence of the internet in traditional political campaigns over the past few cycles (Foot & Schneider 2006; Howard 2006), it is no surprise that the internet came to be an important part of the nomination battles fought in 2005. Perhaps the first clue that the internet would play a unique role in the latest round of Supreme Court appointments came when SCOTUSblog (http://www.scotusblog.com/), a blog devoted to discussion of Supreme Court jurisprudence, created a separate blog space solely devoted to nominations (http://www.sctnomination.com/blog/). Additionally, the conservative blog RedState (http://www.redstate.com/) had also created a site devoted to judicial nominations (http://www.confirmthem.com/), and liberal groups also began to use the web as a place to discuss confirmed and unconfirmed reports of possible nominees for the high court positions. In addition to sites dedicated exclusively to judicial nomination discussion, diffuse content related to the same events rapidly proliferated within the general political blogosphere. Thus the even before the controversies surrounding the replacement of Justices O’Connor and Rhenquist had begun in earnest, it was clear that future nomination battles would offer a unique opportunity to examine how the public’s experience and discussion of a distinct type of political media event may have been altered by the rise of new media in general and blogs in particular.
Of the three nomination battles that took place in 2005, we chose to analyze discussion of the Alito nomination because it was by far the most contentious, generating a significant amount of political discussion and mobilization by both liberals and conservatives. The first nomination was that of John Roberts, who was originally nominated to replace O’Connor but, in part due to his wide base of support, was not only handily confirmed but eventually moved into Rhenquist’s seat as Chief Justice. The Roberts nomination generated significant media attention, but due to his general appeal among Democrats and Republicans alike (and by implication a lack of dramatic conflict), this nomination did not result in significant political mobilization and discussion. In stark contrast, the nomination of Harriet Miers proved relatively unpopular even among some conservative groups, and lasted just over three weeks before she withdrew herself from consideration. Although Miers is very likely the first Supreme Court nominee to have become the subject of a humorous satirical blog (http://harrietmiers.blogspot.com/), aside from this artifact her nomination generated relatively little in the way of widespread discussion and debate in either the press or the blogosphere.
The Alito nomination, however, had a variety of attributes that made it an ideal case study for mediated deliberation. As one of the most experienced nominees in history, with an Ivy League education and a relatively transparent conservative bona fide, Alito enjoyed consistent support from Republican Senators and conservative citizen groups. At the same time, his reputation as a conservative and his relatively public stance on abortion drew the ire of Democratic Senators and liberal groups. Ultimately, these factors contributed to a dramatic and contentious nomination battle that lasted just over 100 days, from the initial announcement by President Bush on October 31st to the final confirmation vote on January 31st, and featured lively discussion in both traditional mass media as well as within the blogosphere.
Theorizing the Impact of Blogs on Mediated Deliberation
Given the relative paucity of social scientific research focusing exclusively on the role played by blogs in political communication, we approach questions concerning their possible impact on processes of mediated deliberation as special cases of issues surrounding the broader effects of new media on the public sphere. Leaving aside normative judgments that invite familiar optimism-pessimism debates for the moment (although we will return to these in the conclusion), we focus in particular on how internet media differ from traditional news outlets as a way to leverage questions of how blogging may be changing the process by which members of the general public experience mediated discussion of the political events of the day. In an important review essay, Dahlgren suggests that we might best theorize the effects of new media on political communication by focusing on distinct analytic dimensions of the public sphere (2005). Using this framework, we identify ways that unique elements of the web as a medium for political communication, as well as particular aspects of the blog genre of websites, may lead to patterns of mediated deliberation distinctly different from those identified in prior case studies.
The first dimension in which we might look for changes is structural, dealing with the “formal institutional features” of the public sphere (Dahlgren 2005, p. 148). Prior to the rise of new media in politics, the dominant considerations in this regard, as evident in Page’s (1996) original case studies, deal with the political economy and organizational norms of mass media news outlets. Applying this analytic concept to the question of blogs and mediated deliberation, the corollary considerations involve the structural differences between the production of news and public affairs content in the blogosphere, as compared to traditional news outlets. In economic terms, the most obvious feature of the web in general and weblogs in particular is the dramatically lower cost structure involved in producing content (Bimber 2003; Drezner & Farrell 2004). Indeed, with the advent of Blogger.com’s simple service, and a variety of others like it, setting up a blog and posting content is something that is within reach of just about anyone with access to the internet, with very little effort. Beyond the economics of time and resources, it is also clear that one of the defining elements of blogs is the relative freedom from the norms and conventions of journalistic practice and objectivity that bloggers enjoy as compared to traditional journalists. Finally, since they are their own publishers, bloggers are largely free from editorial constraints on sourcing, verification, or traditional definitions of “what’s news” (Gans 2004).
Although it is likely too soon to tell, we suspect that these structural differences may lead to the emergence of novel and different norms, regularities, and logics of appropriateness regarding the production of blog content. Just as the classic sociological studies of news production documented a relationship between structural characteristics of American news outlets and patterns of news production, we believe a similar process may be currently unfolding for political blogging. Taking these factors into consideration, we contend that the blogosphere as a whole may be expected to differ structurally from the traditional system of mediated deliberation in terms of the volume of coverage offered on various issues, and the level of attention to distinct elements of those issues. Applied to the case of mediated deliberation over the Alito nomination, these considerations inform our first research question.
RQ1: To what extent do blogs differ from traditional news media in their patterns of attention to and coverage of the Alito nomination battle?
The second analytic dimension of the public sphere is representational, and lies directly at the heart of analyses of mediated deliberation in that it involves the completeness with which mediated discussions stand in for the range of all possible discussion points that may be of interest to members of the public. Indeed, with respect to the representational dimension, “one can raise all of the familiar questions,” including those dealing with “fairness, accuracy, completeness, pluralism of views, agenda setting, ideological tendencies, modes of address, and so forth” (Dahlgren 2005, p. 149). In the case of mediated deliberation in the blogosphere, the most relevant consideration is clearly the ideological and partisan tenor of the sphere as a whole. In part based on the structural characteristics just identified, the blogosphere is a notoriously partisan ideological space. Although a wide variety of ideological positions are represented through blogs, including centrist and hybrid perspectives, it is clear that the bulk of blogs may be easily categorized with familiar ideological labels such as conservative and liberal (Hargittai 2005). Furthermore, as scholars such as Cass Sunstein (2001) have argued, the willingness and ability of internet users to personalize their media experience through ideological filtering raises important questions concerning the extent to which blogs might help to facilitate ideological balkanization and fragmentation in online mediated deliberation.
In any public deliberation, certain themes and developments tend to privilege one side or the other of prevailing discussions much in the same way that political parties can be said to “own” certain issues (Petrocik 1996). In campaigns or political conflicts that resemble them, this creates disincentives for actors to engage on the same issues (Simon 2002), as well as positive incentives to engage in what has been termed “heresthetic,” or the art of persuasion through carefully structuring the alternatives presented to an audience (Riker 1996). Freed from the constraints of journalistic objectivity and fairness, it is thus entirely reasonable to expect that partisan or ideological bloggers may engage in strategic framing of discussions in order to favor their own perspectives. Thus, something akin to the “structured debate” Page identified in the New York Times’ coverage of the first Iraq war may be reasonably predicted in the blogosphere, although likely in an even more polarized form given the dramatic differences in expectations for objectivity and fairness (1996). Applied to the case of the Alito nomination, we formulate our second research question on the basis of these considerations.
RQ2: To what extent do liberal and conservative blogs engage in selective discussions of issues related to the Alito nomination battle?
A third dimension of the public sphere in which we may expect to see differences introduced by the unique nature of blogs as a medium of political communication is interactional. Focusing on the communicative aspects of the public sphere, Dahlgren identifies two subcomponents of the interactive dimension: interaction with media (using and navigating media content to make sense of political events) and interaction with others (through conversation). Naturally, one of the defining characteristics of online communication is its highly interactive nature (Bucy 2004; DiMaggio et al. 2001). With blogs in particular, we identify two interactive possibilities relevant to processes of mediated deliberation. First, to the extent that hyperlinking itself is a form of interactivity (Sundar et al. 2003), we may consider links from blog posts to external content as interactive, especially if the content is produced by other actors traditionally dominant in processes of mediated deliberation. Second, we may further look to the presence of comment functions on many blogs that enable users to directly engage in discussion with blog authors and other readers concerning political events. Both of these unique features of the blog medium relate directly to the two elements of public sphere interaction identified by Dahlgren respectively – links facilitate a more interactive relationship with media content itself, while comments create the potential for conversations to emerge within each individual blog post. Thus our third research question concerns the sense in which blogs may serve to enhance the interactive dimension of mediated deliberation within the public sphere.
RQ3: To what extent to do blogs enhance the opportunities for interactive engagement and discussion surrounding the Alito nomination?
Data and Methods
To explore these questions, we devised a unique research design that enabled us to conduct a specific and detailed analysis of the role played by political blogs in processes of mediated deliberation surrounding the Alito nomination. Given the vast amounts of media coverage devoted to Alito in the fall of 2005, and the even more vast (and difficult to catalogue) terrain of the blogosphere, we chose to limit our analyses to coverage of Alito in the New York Times, and to blog posts featuring links to stories about Alito appearing in the Times. In this way, following extant studies of mediated deliberation (e.g. Bennett et al. 2004; Page 1996; Simon & Xenos 2000), we allow the Times to represent mainstream newspaper coverage and to also serve as the reference point for our investigation of blog coverage, literally treating it as a ‘paper of record.’ Using the Lexis-Nexis database, we began by identifying all Times stories featuring “Alito” in the headline or lead paragraphs, appearing between October 29th 2005 to February 7th 2006. These parameters enabled us to develop a comprehensive corpus of traditional media coverage of Alito’s nomination spanning the days just prior to Bush’s announcement (during which there was a great deal of speculation over the identity of the final nominee) through the week following approval by the Senate. To capture blog discussion, we used an early incarnation of the Blogrunner website (also known as the Annotated New York Times), which provided an automated accounting of all blog posts featuring links to stories appearing in the New York Times. All stories from the Times and all blog posts linking to Times coverage, including the Blogrunner pages used to identify them, were then archived and subjected to content analysis. This created two data sets, the first consisting of all stories about the Alito nomination produced by the Times, and the second consisting of all blog postings identified by Blogrunner as featuring a hyperlink to those stories, with both temporally bounded by the dates mentioned above.
New York Times Stories
Our Lexis-Nexis search of Times stories featuring “Alito” in the headline or lead paragraphs between October 29th 2005 and February 7th 2006 yielded a total of 264 stories. Filtering out duplicate stories, unsigned editorials, and items that fell below a minimum threshold of 325 words (mainly corrections and simple photo captions) reduced the corpus to 153 stories.
All data from the Times coverage was gathered using computer-assisted techniques. At the most basic level, we catalogued the title, date, byline, and number of words for each story, as well as whether the story was an editorial or news item. (Since our principal interest was mediated deliberation in the blogosphere, we did not gather data on where the stories may have appeared in the paper editions of the New York Times.) Beyond these basic data, we also developed a system for identifying sub-topics of the Alito story based on a word-frequency analysis of the coverage as well as our own readings of newspaper and blog discussion. On the basis of these explorations, we identified seven sub-topic categories relevant to mediated deliberation on the Alito nomination. These categories included Abortion, Executive Power (including issues surrounding domestic surveillance), a possible Filibuster, Alito’s Past Record (as linked to materials released in association with his application for the position of Assistant Attorney General in the Reagan Administration), Police Searches (as related to Alito’s prior decisions in cases involving claims of unreasonable search), Judicial Restraint, and Qualifications (e.g. references to Alito’s education and experience). These categories were then used to create a specialized dictionary using the WordStat module of the SimStat statistical package, from which binary measures of topic mentions for each story were generated. Table 1 provides a summary of the proportion of stories mentioning each of these sub-topics.
Note:a Cell entries represent the proportion of stories mentioning each topic. Categories are not mutually exclusive. N= 153.
As mentioned earlier, we gathered blog posts using a tool called Blogrunner, which tracked blog posts linking to Times content. Blogrunner worked in a manner very similar to the “Who’s Blogging” feature found on the web version of the Washington Post. Such tools enable readers of online newspapers to automatically click from a given story to a listing of blog postings that link to the story he or she has just read. Specifically, for each story in our Times database, we located the relevant page within the Blogrunner site containing the list of blogs linking to that particular story. Ultimately, this process yielded 654 posts appearing in 208 unique blogs, which were then archived using the software application TelePort Pro. Although the vast majority of these blogs would easily fit definitions of “political blogs” used by other researchers (e.g. Hargittai 2005), consistent with wide diversity among the types of blogs available online, others would not. However, for our limited purposes, we will consider each blog as political in the sense that it featured commentary on one of the most significant political stories of late 2005.
In coding the blog posts, we used human coders to tap the following characteristics of each blog and individual post. Each of the 208 individual blogs in the list was first categorized as either “liberal,”“conservative,” or “moderate/other.” A random subset of 20 blogs was double-coded for reliability testing. These tests were conducted using Krippendorf’s alpha, which accounts for the possibilities of inter-coder agreement by chance, and produced an alpha of .82. In addition, each post was then coded for its length (using a simple word-count tool), the number of links to other blogs included in the post, the number of links to other press sites (besides the New York Times), and the number of comments to the post (if available). Naturally, reliability figures for the easily gathered data (number of words, number of comments), were nearly perfect. Krippendorf’s alphas for the number of blog links and press links respectively were .71 and .95.
Considerations Regarding the Data
To be sure, in some ways the approach to collecting data pursued here is unconventional and limited in important ways. One limitation is that our print coverage dataset, organized around New York Times coverage only, excludes other forms of mass media such as television and other print sources. A second concern arises from the blog posting data set, in that it excludes discussion of Alito in the blogosphere that for whatever reason does not include a link to coverage by the New York Times. Specifically, given popular perceptions concerning the readership of the Times, one may reasonably wonder whether the blogs included in our study over-represent liberal bloggers. Taking these concerns into account, however, along with the costs and benefits of various alternative strategies, we opted for the present design on the basis of three principal considerations, beyond the simple accessibility of the data.
First, we contend that the data gathered are broadly representative of discourse in both traditional news media, as well as the political blogosphere. Regarding print media, as mentioned earlier our choice of the Times coverage was based on the model of earlier studies of mediated deliberation. As an agenda setter in the old media world, we believe that the Times coverage provides a useful proxy for the general tenor and content of broader national news media. In terms of the representativeness of our blog data, a number of characteristics of the blogs included in the analysis speak against a disproportionate presence of liberal voices. For example, of the over two hundred unique blogs analyzed, a significant proportion was identified as consisting of blogs offering a principally conservative perspective on political affairs. Further, looking at the spectrum of ideological representation from left to right, we find that the ideological spread as a whole mirrors that observed by researchers using more elaborate random sampling techniques. For example, whereas Hargittai (2005) found a conservative-liberal split of roughly 38% to 42% among the political blogs included in their more elaborate random sample, the comparable figures within our data are 35% conservative and 46% liberal. Moreover, by drawing our data from a bounded set of actual hyperlinks (rather than attempting to sample from a “universe” of blogs assembled from a directory) we avoid problems associated with the constantly shifting population of blogs at any one moment, which make identification of a true universe from which to sample nearly impossible. Finally, this method also results in a dataset that includes representative examples of a wide variety of political blogs, with a wide variety of readerships from so-called A-List blogs like the Daily Kos (http://www.dailykos.com/), to smaller and lesser-known blogs with more limited readerships.
Second, by focusing on concrete linkages to traditional media content, we believe the blog posts represented in our data speak specifically to unique patterns of mediated deliberation in a new media context. They often include blog commentary on the events of the Alito nomination, as well as about coverage of the nomination itself in the Times. Moreover, they represent a subset of the blogosphere that is especially visible to online news consumers who might use a feature such as Blogrunner or the Who’s Blogging tool fount on the Washington Post website. By design they explicitly involve instances of new media working in tandem with a traditional media outlet in processes of mediated deliberation. Thus, to the extent that some blogs commenting on the Alito nomination in general (though not featuring a hyperlink to Times coverage) are not included in our analyses, we may also reasonably conclude that such blogs are slightly more peripheral to the processes of mediated deliberation we are interested in.
Finally, the design is especially suited for analysis of selective commentary along ideological lines. By categorizing blogs along the ideological spectrum, and documenting the topical contents of each Times story included in the analysis, the design enables us to determine whether certain kinds of stories are more or less likely to attract links to, and comments from, certain kinds of blogs. If we imagine the blogosphere as a collection of resources available to individual citizens interested in an additional level of distillation and synthesis beyond that stipulated by traditional models of mediated deliberation, such patterns might help to illuminate questions about what facts or events become relevant to them and what kinds of information they are likely to consider in informal conversations or formation of individual opinions.
To address the first research question, concerning patterns of coverage, we divided the dataset into 15 one-week periods, and aggregated the data using Week as the break variable. We then calculated the total number of stories appearing in the New York Times and related blog posts appearing in each of the 15 weeks. The assumption here was that if blog commentary tightly followed Times coverage, we might conclude that the blogosphere was following the agenda set by the traditional media. However, to the extent that bloggers show patterns of interest or coverage that diverge from the volume of Times coverage over each of the weeks, we may interpret that as evidence of structural differences between online mediated deliberation and mediated deliberation through the traditional channel of mass media outlets like the Times. Figure 1 depicts both sets of figures over time, and reveals that while the overall shape of the two line graphs is similar, there are key points at which the blogosphere appears to take up the news of the week with greater vigor than the the Times. These points of divergence are more clearly visible in Figure 2, which plots the ratio of blog posts to Times stories for each week.
Recalling that there are roughly four times as many blog posts as news stories overall, we can interpret data points in Figure 2 as indicating a rough parity of discourse when near 4.0, but indicating divergences as we move above or below this 4-to-1 reference point. Clearly, the data points for weeks six, eight, and thirteen stand out, as do the low points near weeks seven, and nine through twelve. At these points we may see examples of either the blogosphere serving as a forum for intensive discussion of items identified in the Times coverage, or examples of events that carry more weight for a traditional media outlet than for political blogs. Thus with respect to RQ1, the data suggest that the blogosphere does indeed appear to operate relatively independently from traditional news media, possibly reflecting a significant structural change (though likely far short of a “transformation”) in the public sphere. Although the relatively small number of overall stories and posts might lead one to use caution in drawing strong conclusions from these patterns, it is interesting to note that the two highest data points (weeks eight and thirteen) in Figure 2 correspond with the release of archival documents related to Alito’s Assistant Attorney General job application (recall that these documents included a number of politically explosive statements), and widespread discussion of a possible filibuster in the Senate, respectively. Alternatively, in week eleven, the peak of Times coverage coincides with the Senate hearings, an obvious source of many event-driven stories for traditional mass media.
In order to shed light on our second research question, concerning ideological selectivity and representation within the blogosphere, we returned to the disaggregated data, regressing the number of blog posts associated with each story on a series of dummy variables representing mentions in each of our topic categories, along with the Editorial and Story Words variables entered as controls. In generating our dependent variables, each blog post was counted only once, regardless of whether or not the post included multiple hyperlinks to a given story. On the assumption that bloggers of a particular ideological bent might behave differently, we estimated models using the total number of blog posts associated with each story, as well as variables representing totals from each of our ideological categories. Essentially the models attempt to explain variation in the amount of mediated discussion generated by individual stories among liberal, independent, or conservative bloggers based on characteristics of those stories. The logic behind this analysis was that significant coefficients for various story characteristics (especially thematic features) would indicate the extent to which a particular segment of the blogosphere may have exhibited a preference for commenting on stories featuring mentions of specific sub-topics. To the extent that liberal and conservative blogs show preference for different kinds of stories within the story database, we may interpret this as evidence of ideological fragmentation along the lines suggested by Sunstein (2001) and others. Alternatively, to the extent that liberal and conservative blogs show preference for the same topic areas, or to the extent that the blogosphere as a whole shows preference for certain areas, we may interpret this as evidence of healthy (or at least balanced) mediated deliberation on these topics. A summary of all four of the regression models we tested is found in Table 2.
Table 2. Blog Posts as a Function of Topic Mentions
Notes:a Dummy coded variable of whether or not the Times story is an editorial (1 = editorial).
Number of words in the Times story.
N= 117. Cell entries are unstandardized coefficients with standard errors in parentheses.
Examining the coefficients in each of the four regression models suggests that the balance of evidence in the Alito case suggests that blogs contribute to ideological balkanization within processes of mediated deliberation. On the positive side, it appears that both liberal and conservative bloggers were drawn to discussion of the possibility of a filibuster of the Alito confirmation vote in the Senate. However, based on prior research examining on- and off-line political communication in campaign environments, we suspect that even in these instances the dialogue is likely colored by strategic framing and represents an “indirect” form of dialogue at best (Bennett et al. 2004; Simon 2002; Xenos & Foot 2005). Additionally, a number of other findings in these analyses point more clearly to patterns of more overt selectivity in discussions of the sub-issues surrounding the Alito confirmation battle among liberal and conservative bloggers. For example, liberals appear significantly more likely to comment on stories featuring mentions of issues surrounding executive power, undoubtedly fueled by the then-current controversy surrounding President Bush’s domestic wiretapping policies, and the role of the judiciary in checking executive power. Additionally, they also appeared drawn to issues surrounding Alito’s decisions in police search cases (though this is only marginally significant). The significant coefficient for editorials further suggests that liberal bloggers were much more likely to engage in sharply partisan commentary on the Alito nomination, as compared to moderate and conservative bloggers. Turning to the fourth column of Table 2, we see a marginally significant result for conservative attraction to discussion of abortion issues, along with a significant negative coefficient for stories with direct mention of Alito’s past record, as illuminated by documents related to his 1985 application to the Attorney General’s office, which included among other things strong statements about abortion rights, as well as evidence of Alito’s membership in a controversial Princeton Alumni Group with strong anti affirmative-action positions. On the whole then, the analysis suggests that in the case of mediated deliberation surrounding the Alito nomination, the blogosphere served to further divide the representational dimension of the public sphere, diminishing the possibilities for shared reference points in broader public discussions.
To address our third research question, we generated a series of descriptive statistics related to the interactive features of blog posts included in our coding analyses. Given the nature of the data available, we were not able to ground these analyses in direct comparisons between the contributions of blog and Times coverage of the Alito controversy. However, we believe these descriptive indicators reveal a number of things about the unique contributions of blogging to the interactive dimension of the public sphere and mediated deliberation. For example, an informal survey of the blogs included in our study revealed that while a non-trivial proportion of them were produced by legal scholars (e.g., law professors like Eugene Volokh, author of Volokh Conspiracy, http://volokh.com), consistent with research on the demographic characteristics of bloggers (Drezner & Farrell, 2004), most could be classified as ordinary citizens, taking advantage of new media affordances for engaging in public debate over matters of political concern, and not representing themselves to readers as experts. Moreover, our content analyses revealed a highly interactive space within the blog posts, with the mean number of links to other blogs at 1.5 and the mean number of links to other press sources at 5.46 (N= 654). Thus the average post in our database includes reference to at least one other blog (although this was inclusive of links to other posts within the same blog), and nearly a half-dozen links to other media sources. Additionally, among the 387 blog posts in the database located on sites that include a comment feature, the mean number of comments was 16.77; of the 208 total blogs included in our study, nearly 80% (166) offered commenting as a feature, which is nearly twice the proportion estimated for the blogosphere as a whole (Herring et al. 2004). Though unsurprising, these data serve to document the rich contribution of blogs to the interactional dimension of mediated deliberation within the public sphere.
The findings presented here offer a unique window into important changes in the nature of the public sphere that are the result of new media forms. As with any set of case study findings, the results presented here are limited in their ability to provide us with widely generalizeable claims concerning the nature of mediated deliberation in an era of new media. Without examining a broader range of issues and media, it would be difficult to extrapolate these findings to processes of mediated deliberation as a whole. However, they do offer a detailed picture of the ways in which new media may change processes of mediated deliberation by making available a new set of materials that contribute information to the mechanisms of aggregate opinion formation. Though not conclusive, we believe the patterns revealed here are suggestive of important changes in the public sphere and highlight important dynamics of new mediated deliberation that deserve further theoretical and empirical attention.
One of the most important changes in the public sphere suggested in these data concerns the variations in how the traditional news media and the blogosphere react to political events such as the Alito nomination. Taken together, we believe the findings reported here concerning our first research question, as well as some of those related to the second, suggest unique characteristics in the structural characteristics of the blogosphere as a key element of mediated deliberation. As mentioned earlier, we believe that the coverage pattern of the Times largely conforms to patterns of “news values” identified in classic treatments of the structural characteristics of modern news media (Bennett 2003; Gans 2004; Tuchman 1978). Recall that the two spikes in Times coverage by week coincide with the initial announcements and speculations surrounding the Alito nomination, as well as the Senate hearings. Both of these sets of events are notably rich in the classic earmarks of traditional news leads, especially the hearings, which featured drama, a variety of official commentary, and perhaps most importantly, a set of high-profile events.
In contrast, we believe that whereas the traditional news exhibits a highly event-driven structure (Lawrence, 1996), the blogosphere appears to exhibit an information-driven structure. It is notable, for example, that one of the largest spikes in the post-to-story trendline corresponds to the initial release of the archival memos and materials concerning Alito’s prior application to the Attorney General’s office. Although the presence of references to these materials was not positively associated with blog commentary, many of the substantive topics to which the materials were relevant were. Indeed, as websites such as The Smoking Gun (http://www.thesmokinggun.com) illustrate, the ability to base commentary on official documents (such as an application for a job with the Federal Government) provides a powerful formula for weblog content production. In this sense we believe the findings presented here suggest the need for future research on the dynamics of “blog values” (a corollary to “news values”), informed by the theoretical literature on new media, but guided by the classic approach to news values research first demonstrated by Tuchman (1978) and Gans (2004) in prior decades, and elaborated more recently by Bennett (2003).
As second dynamic of new mediated deliberation revealed by our data concerns the fragmented nature of discussion in the blogosphere suggested by the regression analyses and the interactive nature of the blogosphere suggested by our descriptive data on links and comments. Consistent with the broader literature on new media and politics, this finding provides an added and illustrative example of the potential for political polarization and balkanization in new media. Assuming liberals and conservatives filter their blog consumption along ideological lines, the results reported here suggest that those of opposing ideological outlooks may have come away from mediated deliberations over Alito with distinctly different understandings of the key focal points of discussion and relevant issues of the debate. This finding highlights the need for future research that explores the extent to which this assumption holds, specifically concerning the frequency with which hyperlinks cross ideological divides in the blogosphere and the likelihood of individual consumers to follow them (e.g. Hargittai 2005). If the link structure of the blogosphere does indeed create the conditions for “enclave deliberation,” then the content and comment patterns revealed here suggest a great potential for destructive patterns of group polarization in the new mediated deliberation environment (Sunstein, 2001).
On the whole, we believe the findings presented here provide reasons to be both optimistic as well as pessimistic about the contribution of blogs to processes of mediated deliberation in the U.S. On the one hand, the information-driven character of blog coverage and discussion, as well as the prevalence of non-experts in blog posting and commenting, suggest important advances in mediated deliberation over the patterns and problems identified by Page’s early studies (1996). However, the data also provide a detailed image of the potential for new media forms to introduce greater fragmentation and polarization into the public sphere, diminishing the chances for effective deliberation. It is hoped, however, that future research on the precise nature of these dynamics will lead to greater understanding of these processes which remain so vital for vibrant democratic governance.
About the Author
Michael Xenos is an assistant professor in the Department of Communication Arts at the University of Wisconsin-Madison. His research explores the effects of new media forms on civic engagement and democratic citizenship.Address: Department of Communication Arts, University of Wisconsin-Madison, 6136 Vilas Communication Hall, 821 University Avenue, Madison, WI 53706