Political Blogs and the Bloggers Who Blog Them:Is the Political Blogosphere and Echo Chamber?
by Kevin WallstenDepartment of Political Science
University of California, BerkeleyPaper Presented at the American Political Science Association Annual Meeting
Washington, D.C.September 1-4, 2005
Abstract
Most studies of political blogging have focused exclusively on the so-called “A-list” political blogs. While these studies have provided important insights into the content of A-list political blogs (Adamic and Glance, 2005), how A-list political blogs influence media coverage (Drezner and Farrell, 2004) and who A-list political bloggers are (McKenna and Pole, 2004), they have largely ignored the thousands of less read political blogs that are written by average citizens every day. As a result, relatively little is known about political blogging “by the rest of us” (Schiano et al., 2004). Perhaps most surprising in this regard is the fact that while there has been much debate over whether political blogging is a form of political participation (McKenna and Pole, 2004), there has been no systematic research into how ordinary people are using blogs as a form of political expression. In this paper, I will address this oversight by using a computer assisted, quantitative content analysis of 25 randomly selected, non-A-list political blogs over the six month period from July to November 2004 in order to determine the relationship between mainstream media coverage and political blog discussion. The results of this study will help shed light on whether the political blogosphere is a merely an “echo chamber” for the messages of political elites as reported in the mainstream media.
1
Introduction
Although the term “blog” (which refers to a regularly updated personal web site with
posts appearing in reverse chronological order)1 has been around since 1997, political blogging
seems to be a more recent phenomenon. According to most accounts of the history of blogging, 2
politically oriented blogs first emerged after the events of September 11, 2001 – as people turned
to blogs to both express their feelings about the terrorist attacks on America and to locate
information not available in the mainstream media (Scott, 2004). In 2002, many of the bloggers
who were turned on to blogging as a result of September 11th began blogging about the potential
for a war with Iraq. Some of these political bloggers, such as Instapundit, Daily Kos, and
AndrewSullivan.com, began to attract wide readerships as a result of their discussions of
American domestic and foreign policy. In fact, some of these so-called “A-list” blogs (i.e. blogs
that receive a high number of “hits” and incoming links) became so popular that many political
elites and traditional media institutions began to adopt the form themselves. In fact, by 2004
blogging had become so mainstream that all of the major presidential candidates – ranging from
Howard Dean to George Bush – had started their own blogs and both of the major political
parties had granted press credentials to political bloggers to cover their conventions.3 Some
evidence suggests that the mass public has been taken in by all of this high profile blogging.
According to a recent study by the Pew Internet & American Life Project,4 for example, 32
1 There are countless definitions of what a blog is. According to Drezner and Farrell (2004), for example, a blog is “a web page with minimal to no external editing, providing on-line commentary, periodically updated and presented in reverse chronological order, with hyperlinks to other online sources.” According to Mortensen and Walker (2002) blogs are “frequently updated websites, usually personal, with commentary and links.” On her blog, “The Professional-Lurker,” Lois Ann Scheidt defines a blog as, “A weblog is a frequently modified website that allows updating with items that are grouped primarily by the time and/or date of posting. Entries usually appear in reverse chronological order. Contents of the weblog may be available publicly or through restricted access. Weblogs may also utilize special software designed for this implementation.” McKenna and Pole (2004) define blogs as "web pages that are regularly updated, operated by one person or a small group of people and developed for little to no cost. Individual entries or posts are chronologically updated like a diary with the most recent posts viewed first."2 For a good history of weblogs see Blood (2000) and Blood (2003). 3 The credentialed bloggers posted their entries at www.conventionbloggers.com for the Republican National Convention and at www.cyberjournalist.net for the Democratic National Convention. 4 For more details on this study see http://www.pewinternet.org/pdfs/PIP blogging data.pdf.
2
million U.S. citizens now regularly turn to weblogs for information and 9% of Internet users said
they read political blogs “frequently” or “sometimes” during the 2004 presidential campaign.5
This explosion in the popularity of blogging has been accompanied by a surge in the
number of academic studies focusing on blogs. Drezner and Farrell (2004), for example, have
found that journalists are increasingly likely to refer to “A-list” political blogs as part of their
news gathering routines and that these “A-list” blogs receive a disproportionate number of links
from other blogs. In a somewhat different vein, McKenna and Pole (2004) surveyed the authors
of the most “popular” political blogs in order to determine who blogs, how popular bloggers use
their blogs and what impact political blogging has on traditional forms of political participation.6
More recently, Cornfield et al (2005) have attempted to link the content of “A-list” political
blogs to mass media coverage by tracking the discussion of a number of high profile political
issues during one month of the 2004 election campaign and Adamic and Glance (2005) have
analyzed the posts of 40 “A-list” political blogs during the two months prior to the 2004
Presidential election in order to determine which issues and actors were most frequently
discussed. In short, there has been no dearth of studies focusing on political blogs.
While all of these studies about political blogs have provided important insights into the
dynamics and consequences of “A-list” blogs, they have largely ignored the thousands of less
read political blogs that are written by average citizens every day. As a result, relatively little is
known about political blogging “by the rest of us” (Schiano et al., 2004). Perhaps most surprising
in this regard is the fact that while there has been much debate over whether political blogging is
5 There is also indirect evidence that Americans turned to political blogs with increasing frequency during the 2004 Presidential campaign. For instance, the most popular political blogs, such as DailyKos, Wonkette and Instapundit, received dramatic increases in the level of “traffic” in the months leading up to the 2004 Presidential election. See http://techcentralstation.com/011105B.html.6 Popular here refers to how widely read the blogger’s blog is. In order to determine who the popular bloggers were, McKenna and Pole selected the top 125 political weblogs using the rankings from four websites that rank the popularity of blogs: Blogstreet, the Truth Laid Bear Ecosystem, Technorati’s Top 100 and the Truth Laid Bear. Each of these sites has slightly different ways to measure popularity – ranging from the number of incoming links to the number of unique “hits.” In order to account for these differences, McKenna and Pole decided to include only those weblogs that appeared on at least two of the four top rankings.
3
a form of political participation, there has been no systematic research into how ordinary people
are using blogs as a form of political expression. In this paper, I will address this oversight by
using a computer assisted, quantitative content analysis of 25 randomly selected, non-A-list
political blogs over the six month period from July to November 2004 in order to determine the
relationship between mainstream media coverage and political blog discussion during a period of
high political engagement – the presidential election campaign. More specifically, this paper will
test the hypothesis that the political blogosphere is no more than an “echo chamber” for the
messages of the mainstream media.
Political Blogs: Participation or Expression?
With so much political blogging going on, there has been a rush to determine where
political blogging fits into traditional classifications of political activity. In particular, there has
been a great deal of debate over whether political blogging constitutes a new and unique form of
political participation.7 Whether political blogging is, in fact, a form of political participation, of
course, depends primarily on the definition of political participation that one is working with. In
the political science literature, political participation is usually defined as “action directed
explicitly toward influencing the distribution of social goods and social values” (Rosenstone and
Hansen, 1993). Unlike the definitions used by practitioners in other fields, this definition points
to the importance of the individual’s motives in engaging in a particular activity.8 If the
individual is motivated by a desire to influence the distribution of goods and values, then the
activity is participation. Similarly, if they are motivated by some other goal – such as the desire
for self expression – then the activity is not political participation.
In order to know whether political blogging is political participation, therefore, it is first
7 For a good example of this discussion see McKenna and Pole (2004). 8 In their analysis of political blogging, for example, McKenna and Pole (2004) argue that the act of writing and thinking about politics and sharing one’s views with others constitutes political participation. As a result, they claim that blogging needs to be included among other more traditional forms of participation, such as voting, attending rallies and signing petitions.
4
necessary to know what the motivations of political bloggers are and how they use their blogs.
Indeed, if bloggers use their blogs to run political advertisements, encourage their readers to sign
on-line petitions or mobilize their readers to vote, then political blogging is probably best
understood as a form of political participation. If, however, bloggers are using their blogs as
nothing more than technological “soapboxes” from which to spout their opinions into the vast
expanse of cyberspace, an alternative definition of political blogging may be required. Of course,
the appropriate way to uncover the motivations of political bloggers and their uses of blogs is
through a combination of survey data and content analysis. In short, the question of whether
political blogging is, in fact, political participation is an empirical question that, unfortunately,
has not yet been adequately addressed in the literature.9
More important for my argument in this paper than whether political blogging is a form
of political participation, however, is the recognition that political blogging is, at its core, an
action designed to allow individuals to express their opinions on political issues. Put differently,
regardless of whether political blogging is a form of political participation, political blogging
may be best defined as a novel form of political expression. The argument to define political
blogs as political expression rather than political participation has more than semantic
consequences. Indeed, defining political blogging as a form of political expression is significant
because it raises a fundamentally different set of questions than defining political blogging as a
form of political participation. If political blogging is defined as political participation, for
example, questions such as “who blogs,” “why do people blog,” and “what impact does blogging
have” take center stage. By contrast, when political blogging is defined as a form of political
expression questions such as “what is the content of the political expression found on political
blogs,” “what impact do political events have on the content of the political expression found on
9 The exception to this, of course, is McKenna and Pole’s study. Unfortunately, their study is limited to A-list bloggers. For a discussion of this question as it relates to ordinary political bloggers see Wallsten (forthcoming).
5
political blogs” and “how does the content of the political expression found on political blogs
change over time?” In this paper, I will attempt to answer this latter set of questions by analyzing
the content of the political expression found in ordinary political blogs.
Is the Political Blogosphere an “Echo Chamber?”
For those who have at least a passing familiarity with the political blogosphere, the claim
that political blogs can be understood primarily as forms of political expression is probably
unsurprising. Indeed, as the number of political bloggers has grown and a collective sense of
self-awareness has emerged within the political blogosphere, political bloggers and academics
alike have begun to ask questions about both the content of the political expression found in
political blogs and about the external factors that influence that content. In addition to the
emergence of a number of internet sites devoted exclusively to measuring the content of blogs –
such as Daypop, BlogPulse and BuzzMetrics – a number of recent studies have focused
exclusively on describing the content of the political expression found on political blogs. In a
study by Adamic and Glance (2005), for example, the number of times specific political actors
and news stories were mentioned by political bloggers was counted over the course of a day and
comparisons were made between liberal and conservative bloggers. Similarly, a new Pew study
conducted by Cornfield et al. (2005) attempted to link discussion by A-list political bloggers to
media coverage by tracking the discussion of a number of high profile political issues during one
month of the election season.
One of the most popular topics in this recent wave of concern over the content of political
blogs has been the question of whether the political blogosphere constitutes an “echo chamber.”
What does it mean to call the political blogosphere an echo chamber? The idea that the political
blogosphere is an “echo chamber” is the fairly straightforward notion that the content of political
blogs does little more than “echo” the messages of put forth by political elites in the mainstream
6
media. Although few have spelled it out in precisely this way, the idea that the political
blogosphere is an echo chamber is really made up of three separate ideas: (1) that the issues
discussed in political blogs will closely mimic the issues discussed in the mainstream media
(such that more media coverage on an issue leads to more discussion of that issue in the political
blogosphere); (2) that the positions taken by political bloggers on political issues greatly
resemble the positions taken by the blogger’s own party leaders and publicized in the mainstream
media; (3) the frames and justifications employed by political bloggers are the same as the
frames and justifications of the blogger’s party leaders as publicized in the mainstream media. In
short, calling the political blogosphere an echo chamber suggests that the political expression
found on political blogs is closely related to, if not entirely derivative of, the content of
mainstream media coverage.
Although there have been no systematic studies of any of these three dimensions of the
echo chamber hypothesis, there are good theoretical reasons to expect that political blogs should
act as echo chambers for the messages presented in the mainstream media. Starting with the
work of V.O. Key and continuing through present day research, the belief that individual
political attitudes are responsive to elite messages presented by the mass media is widely
accepted by political scientists. Iyengar and Kinder (1984), for example, have found that the
level of importance people attach to an issue is a direct function of the level of media coverage
that issue receives. In a somewhat different vein, Zaller (1992) has argued that the actions and
statements of political elites, as reported in the mass media, exert a strong, if not total, influence
on mass public opinion – with party loyalists adopting not only the policy positions but also the
frames and arguments of their respective party leaders once the appropriate cues are sent out
through the mass media. To the extent that active forms of political expression, such as political
blogging, follow the same dynamics as the passive expressions that generated these studies of
7
public opinion (i.e. responding to public opinion surveys), the hypothesis that the political
blogosphere is an echo chamber has a great deal of evidence to support it.
So is the political blogosphere an echo chamber? Although answering this question in its
entirety would require assessing the relationship between blog content and media coverage on
each of the three dimensions discussed above and is, therefore, well beyond the scope of this
paper, I will address one dimension of this question here: what is the relationship between the
amount of coverage an issue receives in the mainstream media and the amount of discussion the
issue receives in the political blogosphere?
The Predictions of the Echo Chamber Hypothesis
According to the echo chamber hypothesis, there should be a strong relationship between
the level of media coverage an issue receives and the amount of discussion on that issue that
takes place in the political blogosphere. To be more precise, the echo chamber hypothesis makes
four predictions. First, more mainstream media coverage of an issue should lead to more posts on
that issue and less media coverage should lead to fewer posts on that issue. In other words, there
should be a positive relationship between the amount of media coverage on an issue and the
number of posts on that issue – meaning both the correlation coefficient (r) and the slope
coefficient of the regression line (b) should be positive. Second, increases and decreases in media
coverage on an issue should have a fairly direct and large impact on the number of posts. Indeed,
according to the echo chamber hypothesis, political bloggers should blog about what they see
when they see it so there should be a fairly strong relationship between the number of stories on
an issue and the number of blog posts mentioning that issue. In other words, in regression
analysis, the slope coefficients (b) for media stories should be closer to 1 than to 0 – meaning
that one story on an issue in the media should produce about one post in a political blog. Third,
the echo chamber hypothesis says that political blogs will not discuss issues that are not covered
8
in the media and, furthermore, political blogs will not ignore issues that are discussed in the
media. In other words, there should be a tight relationship between media coverage and blog
discussion – meaning that that the correlation coefficient (r) and, therefore, R² are high. Finally,
the echo chamber hypothesis predicts that political bloggers will rely primarily on mainstream
media sources for their information. As a result, the primary source of external links should be to
the mainstream media and the number of mainstream media links should remain fairly constant
over time (because the influence of media does not change). To summarize the predictions of
the echo chamber hypothesis:
P1: The correlation coefficient (r) and the slope coefficient (b) between the amount of media coverage on an issue and the number of posts on that issue should be positive.
P2: The slope coefficients (b) for media coverage on an issue should be closer to 1 than to 0.
P3: The correlation coefficient (r) and R² between media coverage and blog posts should be high (i.e. closer to 1 than to 0).
P4: The majority of external links should be to mainstream media sources and the number of links to media sources should remain fairly constant over time.
It is worth mentioning here that the predictions of the echo chamber hypothesis could be
correct at the aggregate level even if political blogs are acting as media watchdogs instead of
echo chambers. Indeed, if a significant number of political bloggers carefully followed media
coverage on a number of high profile issues and frequently criticized the shortcomings of this
coverage, the predictions of the echo chamber hypothesis – a positive relationship between
media coverage and the number of posts, a slope coefficient near 1 and a high R² - would hold.
In this case, however, it would clearly be a mistake to call the political blogosphere an echo
chamber because media messages are being consistently criticized rather than echoed. It is for
this reason that a more detailed content analysis – one that looks not only at the correlation
between the amount of media coverage and the number of blog posts but also at how media
9
messages are discussed by political bloggers – is required. Unfortunately, this type of analysis is
also beyond the scope of this paper. As a result, I will not be able to determine in this paper if a
high R² and a positive relationship are evidence in favor of the echo chamber hypothesis or just
evidence of political blogs acting as media watchdogs.10
Why Study Ordinary Political Blogs?
Before analyzing the relationship between mainstream media messages and ordinary
political blog content it is worth saying a few words about why ordinary political blogs should be
studied at all. Although most studies of political blogs are justified on the basis that the blogs
under study influence some important macrolevel outcomes, such as media coverage or
international politics (Drezner and Farrell, 2004), these studies, as suggested above, focus only
on the most widely read and influential political blogs. The ordinary political blogs that form the
subject matter of this paper, however, are, by definition, likely to be hidden in the vast expanse
of the internet and have an audience size that is asymptotically close to zero. If we are to care
about studying ordinary political blogs, therefore, we must find some justification other than
their potential impact on the external world.
In fact, there are a number of good reasons to study ordinary political blogs even if they
pose no threat of influencing the environment in which they operate. In the interest of brevity, I
will mention only one of these reasons here. As suggested above, ordinary citizens are
increasingly using political blogs as a form of political expression and – possibly – a form of
political participation. As such, research into who blogs, what is blogged about and how blogs
interact with the world around them is needed to answer questions such as: “Does political
10 Consistent with this idea, some have labeled the political blogosphere a “fifth estate” – separate from the “political nobility,” “the knowledge clergy,” the citizenry and the press (Cornfield et al., 2005). Although labeling bloggers a “fifth estate” may be a somewhat grandiose label for such a nascent phenomenon, the title is meant to suggest that political bloggers use their blogs to either discuss issues that have not been covered by the mainstream media or to discuss issues that have been covered in what they perceive to be a biased fashion. Political bloggers, for example, take credit for drawing attention to issues that were initially ignored by the mainstream media – such as Trent Lott’s comments at Strom Thurmond’s birthday celebration and the authenticity of the National Guard documents presented by CBS News.
10
blogging make people more engaged in and informed about politics?”; “Does political blogging
intensify differences in opinion on political issues or diminish them?”; and, “What kinds of
events motivate ordinary people to make active expressions of opinion on political issues?”
Studies that focus exclusively on A-list political bloggers are unlikely to provide meaningful
answers to these questions because A-list political bloggers are likely to be professional
journalists, political activists or some other kind of unique and exceptional individual (McKenna
and Pole, 2004). Indeed, studying only A-list bloggers is likely to lead us to draw some very
faulty conclusions about the who’s, what’s and why’s of this emerging form of political activity.
As a result, more studies of ordinary political blogs are needed.
Methodology
Sampling Ordinary Political Blogs
One of the most serious obstacles to drawing valid conclusions about the content of
ordinary political blogs stems from the fact that no single population list of political blogs exists
and, as a result, there is no way to generate a truly representative sample of political blogs.11
Indeed, the very idea of a “political blog” is troubling because there is no consensus on what
distinguishes political blogs from non-political blogs.
There have been two main suggestions for how to separate political blogs out from the
more numerous non-political blogs. First, some have suggested that a blog can be defined as
political if the amount of political discussion exceeds some (as of yet unspecified) percentage of
the overall content. Although this method has the virtue of drawing clear lines between political
11 Similarly, there are no population lists for the blogosphere on the whole. If there were, one method for generating a random sample of political blogs would be to use these population lists to generate a random sample of general blogs. Each blog could then be checked for whether it was political or not and either throwing the sampled blog out or including it in the study. This process could be repeated until the desired sample size was reached. This method of generating a sample is excellent from both a practical and methodological perspective not only because it samples from the nearly complete population lists of the above mentioned sites but also, and more importantly, because the researcher does not have to generate these population lists on his or her own. Unfortunately, the two sites that had provided extensive – though not complete – lists of the blogs currently on the internet: www.blo.gs or the NITLE Blog Census (www.blogcensus.net) were no longer working at the time of this paper.
11
and non-political blogs, it is also has the drawback of requiring a tremendous amount of effort
simply to determine which blogs should be included in the research. As a result, this method may
be better for determining the overall level of political blogging relative to other types of blogging
than for answering more detailed questions about the specifics of political blogs such as the one
that occupies my attention here.
Second, some have suggested that the best method for determining whether a blog is
political or not is simply to trust the blogger’s own assessment of their blog. There are two
different ways to employ this method: keywords and interviews. The keywords approach to
defining a political blog relies on the fact that many bloggers list their blogs on blog directories
and, more importantly, these blog directories allow bloggers to categorize their blogs using a
number of different keywords. Since these blog directories usually allow for searches based on
these keywords, it can be relatively easy to locate the blogs that are political. By contrast, the
interview method requires contacting bloggers via email and asking them whether they think
their blogs are political. Much like the content approach discussed above, this method would
have the disadvantage of requiring a large sample of bloggers just to find the small number of
blogs that are political. It seems, therefore, that this method would probably be better employed
as an additional check once a blog had been defined as political by one of the other methods
discussed above. Indeed, given that the goal of this paper is to assess the content of political
blogs and not to determine the level of political discussion in the blogosphere on the whole, it
seems that the keyword approach provides the best method for defining a political blog.12
While the keyword method might be the most appropriate way to generate a sample given
my goals in this paper, it is not without its drawbacks. In addition to the fact that this method
requires a great deal of work the part of the researcher to create the population list, it is also
12 This is the approach used by Adamic and Glance (2005). In their article they create a sample of 1000 political blogs by downloading the listings of political blogs from several online weblog directories, including eTalkingHead, BlogCatalog, CampaignLine, and Blogarama.
12
unattractive because the directories provide information about an unrepresentative subset of the
blog population on the whole. Indeed, the blog directories used to generate the population lists
include only those blogs that have been submitted for inclusion by the authors of the blogs
themselves and, as a result, cannot produce anything like a true or complete list of political blogs
from which to sample. Furthermore, many bloggers may either intentionally (in order to increase
readership) or unintentionally miscategorize their blogs as political when, in fact, political issues
are rarely or never discussed.
Despite these weaknesses, I relied on twelve of the most well known blog directories
(Blogwise, Blogarama, BlogCatalog, BlogUniverse, BlogSearchEngine, BlogStreet,
CampaignLine, ETalkingHead, GetBlogs, EatonwebPortal, Globe of Blogs and Yahoo’s
directory of blogs) to generate a population list of 10,732 unique political blogs. From this list, I
randomly sampled each political blog, checked if the blog was active during the period of the
study (June to November 2004) and either included in the sample or threw it out.13 In all, 83
political blogs were sampled in order to arrive at the final sample of 25 political blogs. Overall,
these 25 political blogs contained 3,763 unique posts over the period of the study. The final list
of blogs included in this study is listed in Table 1. Although the fact that blog directories include
only those blogs that have been submitted for inclusion by the authors of the blogs themselves
and, as a result, cannot produce anything like a true representative sample of political blogs, this
is the best (if not only) method of generating a sample given the goals of this paper.
Content Analysis
For my analysis of the relationship between media coverage and blog content, I chose to
focus on 25 high profile political issues. These 25 issues were selected not only to capture a
broad array of policy concerns – ranging from domestic politics to international politics – but
also, and more importantly, to represent the key political issues being debated during the period
13 Blogs were also checked to determine if they were American.
13
of the study. Although the 25 issues did not capture the total range of issues discussed in political
blogs or in the mainstream media, they did seem to constitute the majority of blog discussion and
media coverage from June to November 2004. A list of these issues and the keywords used to
identify them can be found in the Appendix.
Using the list of issue keywords presented in the Appendix, a “keyword in context”
(KWIC) content analysis program was used to code each of the 3,763 political blog posts in the
sample. Posts were coded for each issue as either having mentioned the issue (i.e. the issue
keyword was used) or not having mentioned the issue (i.e. the issue keyword was not used).
Each mention of an issue keyword was coded for every post and, as a result, many posts are
recorded as mentioning multiple issues. In addition to coding for issues, the number of links to
academic, governmental, mainstream media, partisan media, A-list blogs and liberal and
conservative non-governmental organizations was recorded for each post. Finally, each political
blog was categorized according to the partisanship of its author.14
In order to measure the independent variable for this study – mainstream media coverage
– I used Lexis-Nexis to search the New York Times over the period from June 1, 2004 to
November 30, 2004. I searched the “headlines, lead paragraphs and terms” section of the Lexis-
Nexis archives using all of the issue keywords listed in the Appendix. Every article that
mentioned the keyword in the headline, lead paragraph or terms section was counted for each
day during the period of the study. Editorials and letters to the editor were excluded from the
final counts of the number of stories.
Results
14 In order to code for the partisanship of the blog, three steps were taken. First, the title of the blog was checked for references to partisanship because many political blogs include some reference to their partisanship in their blog title. Only if this was inconclusive were further steps taken to code the blogger’s partisanship. The second step that was taken to determine the partisanship of the blogger was to look for partisan references in the blogger’s description of the blog. The third and final step that was taken was to analyze the content of the blog itself. If a blog made an outright statement about the outcome of the election such as “Please let Bush win,” “Please defeat Bush” or “Vote against Bush,” the blog was coded for the direction of that statement.
14
Before turning to a discussion of the relationship between the amount of media coverage
an issue receives and the number of blog posts on that issue, it is worth saying a few words about
the political blogs that are included in my sample. First, the sample included more Democratic
political bloggers than Republican political bloggers. More specifically, my sample included
fourteen Democratic bloggers, eight Republican bloggers, one independent blogger, one
libertarian blogger and one blogger whose partisanship could not be determined. Although it
might appear that, given my sample, Democrats are over represented in the political blogosphere,
a note of caution is in order. As stated above, this paper looked only at political blogs that were
active from June to November of 2004. It is possible that Democratic bloggers were either more
likely to start their political blogs earlier (making them more likely to be active since June 2004)
or that Democratic bloggers were more likely to continue their blogging through the end of
November 2004. Indeed, if Republican political bloggers were relatively late arrivals on the
political blogging scene (meaning they started their blogs after June 1, 2004) or were more likely
to leave the political blogosphere after the election (meaning they stopped blogging sometime
before November 30, 2004), Democrats would be overrepresented in my sample.
In order to determine if Democrats were overrepresented in my final sample, I coded
each of the political blogs that were excluded from my sample for their partisanship. Once non-
English, non-American and non-blogs were eliminated from the excluded set I was left with a
total of 30 American political blogs (not active during the period from June to November 2004)
to code for partisanship.15 In this set of political blogs, Republicans outnumbered Democrats by a
margin of 14 to 10 (with 3 Independents and 3 unknowns). It does appear, therefore, that
15 Overall, therefore, 25 of the 83 blogs sampled were written in English, American and active from June to November 2004 while 30 of the 83 were written in English, American and not active from June to November 2004. The remaining 28 blogs were either not written in English, not American, did not meet the definition of a blog or did not work.
15
Democrats are slightly overrepresented in my sample because Democrats appear to have been
more likely to maintain an active political blog during the time period I am studying.
Second, the political blogs contained in my sample varied widely in the overall amount of
activity during the time period I am studying. Indeed, while some made as many as 15 posts to
their blogs every day, others updated their blogs only a few times a month. To be more precise,
the average number of posts per day was 1.1 but the standard deviation in the sample was 1.8. It
appears, therefore, that some bloggers are more committed to and invest more time in their
political blogs than others.
Third, the political blogs in my sample varied widely in their overall level of political
discussion. Some of the political blogs in the final sample were exclusively political and offered
almost no details on the personal life of the blogger while other political blogs in the sample
seem to double as personal blogs – with stories and anecdotes about the blogger’s personal life
filling a large portion of the blog. This variation in the amount of political discussion is the
inevitable product of the fact that blogs allow authors to discuss any topic they want and that the
blog directories from which the political blogs in this paper were sampled impose no restrictions
on what keywords can be used to describe one’s blog. As a result, many bloggers may list their
blog as political even if they never discuss political issues at all.16
So how tight is the relationship between the amount of media coverage an issue receives
and the number of posts on that issue in the political blogosphere? In order to fully answer this
question and, in doing so, determine whether the echo chamber hypothesis is accurate, it is useful
to look at the relationship between media coverage and blog posts in three different ways: in the
overall sample, for Democrats and Republicans and for individual political blogs. Analyzing the
strength of the relationship in the overall population is the perhaps the best test for the echo
16 Future studies may correct for this by using the content approach as a further filter for the blogs sampled from the blog directories.
16
chamber hypothesis as I have spelled it out because it primarily makes claims about the political
blogosphere on the whole and not about small subsets of the blogosphere or individual
differences between political bloggers. By looking only at the overall sample, however,
important differences between political partisans and individuals may be missed. More
specifically, the idea that Democrats and Republicans will respond differently to media coverage
on different issues cannot be tested and it will be unclear whether there are some political
bloggers who have assumed the role of echo chamber more than others. In the case of
partisanship, this oversight can be particularly troublesome because, if Democrats and
Republicans are acting as echo chambers for an entirely different set of issues, the overall
correlation between media coverage and these issues will appear small – thereby discounting the
echo chamber hypothesis – even though there is a significant amount of echoing going on. As a
result of these considerations, I will analyze the relationship between issue coverage and blog
posts in the overall sample, in partisan groups and for each individual political blogger.
The findings for the relationship between the amount of media coverage and the number
of blog posts in the overall sample are presented in Table 2. As Table 2 clearly shows, there is no
clear overall relationship between the amount of media coverage of an issue and the number of
political blog posts that mention that issue. Indeed, for some issues there is a significant
relationship between media coverage and blog posts while for others there is no relationship
whatsoever. On the one hand, the relationship between media coverage and blog posts is strong,
positive and statistically significant for 8 of the 25 issues studied: Iraq (r = .25), abortion (r =
.22), prisoner abuse at Abu Ghraib (r = .26), Osama bin Laden (r = .22), the military draft (r =
.21), the crisis in the Sudan (r = .21), moral values (r = .44) and stem cell research (r = .31).
Particularly striking is the relationship between media coverage of moral values and political
blog discussion of moral values. Each additional story on moral values in the mainstream media
17
was predicted to produce an additional .4 blog posts on moral values and, even more impressive,
media coverage of moral values explains nearly 20% of the variance in blogging activity on
moral values. For some issues, therefore, media coverage does seem to exert a strong influence
on blog content.
On the other hand, there was no statistically significant correlation between media
coverage and political blog posts for 17 of the 25 political issues included in this study. For 8 of
these 17 issues (North Korea, immigration, Afghanistan, affirmative action, terrorism, No Child
Left Behind, welfare and the forged documents controversy involving Dan Rather), the amount
of media coverage explained none of the variance in the number of political blog postings.
Furthermore, the slope coefficients of the regression analysis for 16 of these 17 issues were all
small and statistically insignificant. The one exception was media coverage and blog postings on
Iran – where approximately 17 stories are predicted to produce one additional political blog post.
For the majority of issues, however, it appears that there is little relationship between the amount
of media coverage and the number of political blog posts.
In order to explore the possibility that the overall findings obscure important differences
in how partisans respond to mainstream media messages, I also analyzed the relationship
between media coverage and blog posts for Democrats and Republicans. The results of this
analysis are presented in Table 3. As Table 3 clearly displays, there are striking differences
between the strength of the relationship between issue coverage and blog posts for Democrats
and Republicans. While there is, for example, a significant correlation, between media coverage
and blog posts for Democrats on Iraq, the prisoner abuse scandal at Abu Ghraib, gay marriage
and weapons of mass destruction, there is no significant correlation between media coverage and
blog posts for these issues for Republicans. Similarly, while Republicans exhibit a significant
correlation between media coverage and blog posts on abortion, Osama bin Laden, the military
18
draft and the crisis in the Sudan, there is no such correlation on these issues for Democrats. In
fact, the only issues that show a statistically significant correlation for both Democrats and
Republicans are moral values and stem cell research.
To put the differences between Democrats and Republicans into even starker view, it is
worth considering the differences in the slope coefficients for some of these issues. For
Republicans, for example, each additional story on the military draft is predicted to produce a .24
increase in the number of blog posts on the military draft and media coverage on this issue
explains 10% of the overall variation in blog posts. For Democrats, however, the slope
coefficient for the military draft is a statistically insignificant -.03 and media coverage explains
none of the variation in the number of blog posts. On the issue of weapons of mass destruction a
similar divide emerges: eight new stories on weapons of mass destruction in the mainstream
media are predicted to produce one new blog post for Democrats (b = .13) but no new blog posts
for Republicans (b = 0). In short, Republicans and Democrats respond in very different ways to
mainstream media coverage based on the issue being covered.
In addition to differences in how partisans respond to mainstream media coverage, there
may be important differences in the strength of the relationship between media coverage and
blog posts across individuals. In order to explore the possibility that certain individuals are likely
to fit the echo chamber hypothesis better than other individuals, I analyzed the relationship
between media coverage and blog posts separately for each of the 25 political blogs in my
sample. The results of this analysis are presented in Table 4. Analyzing the strength of the
relationship between media coverage and blog posts for each individual blogger reveals that
there are, in fact, important differences between bloggers. More specifically, it seems possible to
identify at least three different kinds of political bloggers based on how responsive the blogger is
to media coverage. First, some political bloggers post in ways that have no relationship
19
whatsoever to mainstream media coverage (meaning that there is no correlation between media
coverage and their blog posts on any issues). In my sample, 11 of the 25 political blogs fell into
this category. As a general rule, it appears that these political bloggers tend to be less prolific in
their posting and these blogs tend to include a significant amount of discussion about the
bloggers personal life in addition to discussion about political issues. Indeed, these political
bloggers posted an average of only .5 posts per day (compared with 1.5 posts per day in the rest
of the sample) and were less likely than the other blogs to mention almost every political issue
keyword.
Second, some political bloggers seem to be responsive to mainstream media coverage
only on one or two issues. To be more precise, seven of the 25 bloggers in my sample displayed
a significant relationship between media coverage and blog posts on one issue and three of the 25
bloggers displayed a significant relationship between media coverage and blog posts on two
issues. In general, these political blogs tend to be updated with greater frequency than the
political blogs that showed no significant correlation between media coverage and blog posts
(with an average of 1.3 posts per day) and are significantly more likely to include mention issue
keywords in their posts.
Finally, a few political blogs showed a strong relationship with media coverage on a
number of different issues. Four of the political blogs in my sample (“2nd Battalion 94th
Artillery,” “Anti-Climacus,” “Buffalo Pundit” and “George in Denver”), displayed significant
correlations between media coverage and blog posts on three or more of the political issues
measured in this study. These political blogs tended to have the highest levels of activity (an
average of 2 posts per day) and were among the highest in the overall percentage of posts
mentioning one of the issue keywords.
20
One final piece of evidence that can be brought to bear on the question of the relationship
between mainstream media coverage and blog posts comes from the links that political bloggers
use in their blogs. According to the echo chamber hypothesis, mainstream media sources should
be the primary source of political information for political bloggers and, as a result, links to
mainstream media sources should be relatively frequent when compared to other sources (such
as other political blogs, non-governmental organizations or partisan publications) and relatively
constant over time. In order to determine which sources are important to political bloggers, I
counted the number of times each political blog linked to academic (.edu domain names),
governmental (.gov domain names), mainstream media,17 partisan media,18 liberal and
conservative “A-list” blogs19 and liberal and conservative organizations.20
Perhaps unsurprisingly, mainstream media sites were by far the most common sources
linked to by political bloggers in my sample. On average, a political blog linked to a mainstream
media source once every five posts whereas an A-list political blog was linked to only once
every 25 posts, a partisan media outlet linked to only once every 50 posts and a partisan
17 I defined the mainstream media as the following organizations: USA Today, the New York Times, Yahoo News, the BBC, the New York Daily News, NBC, the Financial Times, the Associated Press, MSNBC, CBS, ABC, the New York Post, Reuters, the Washington Post, Fox News, the Washington Times, the Christian Science Monitor, the San Jose Mercury News, the Sacramento Bee, CNN, the Boston Herald, the Los Angeles Times, the Chicago Tribune, the Chicago Sun Times, the San Francisco Chronicle, the Seattle Times, the Denver Post, Boston.com, NJ.com and Ohio.com. 18 Links to five conservative and five liberal media sites were measured. The liberal sites were: AirAmericaRadio.com, Michealmoore.com, Motherjones.com, Salon.com and TheNation.com. The conservative sites were: DrudgeReport.com, NationalReview.com, OpinionJournal.com, TownHall.com and WeeklyStandard.com.19 Links to the top 20 liberal and top 20 conservative blogs as listed in Adamic and Glance (2005) were counted. Conservative A-list political blogs included: Allahpundit.com, Andrewsullivan.com, Belmontclub.blogspot.com, Blogsforbush.com, Captainsquartersblog.com, Deanesmay.com, Hughhewitt.com, Indcjournal.com, Instapundit.com, Littlegreenfootballs.com, Michellemalkin.com, Mypetjawa.mu.nu, Powerlineblog.com, Realclearpolitics.com, Rogerlsimon.com, Timblair.spleenville.com, Vodkapundit.com, Volokh.com, Windsofchange.net and Wizbangblog.com. Liberal A-list political blogs included: Americablog.blogspot.com, Atrios.blogspot.com, Blog.johnkerry.com, Crookedtimber.org, Dailykos.com, Digsbysblog.blogspot.com, j-bradford-delong.net, jameswolcott.com, juancole.com, mydd.com, oliverwillis.com, pandagon.net, politicalwire.com, prospect.org/weblog, talkingpointsmemo.com, talkleft.com, theleftcoaster.com, washingtonmonthly.com, wonkette.com and yglesias.typepad.com.20 The liberal organizations were: Aarp.org, Aclu.org, Acorn.org, Aflcio.org, Foe.org, Graypanthers.org, Greenpeaceusa.org, Lwv.org, Moveon.org, Naral.org, Nature.org, Ngltf.org, Now.org, Plannedparenthood.org, Rainbowpush.org and Sierraclub.org. The conservative organizations were: Aim.org, Cato.org, cc.org, frc.org, heritage.org, jbs.org, nra.org and nrlc.org.
21
organization linked to only once every 350 posts. It is important to note, however, that, despite
the relative frequency of mainstream media links, only one in five political blog posts on average
actually linked to a media source and that, when taken together, links to alternative sources of
information comprise a considerable percentage of the overall links provided by political
bloggers. Overall, therefore, mainstream media sources are not an entirely dominant influence in
the political blogosphere.
In addition to not being a dominant influence in the aggregate, it appears that the
influence that mainstream media sources have over the political blogosphere ebbs and flows over
time. Indeed, as Figure 1 clearly shows, the number of links to mainstream media sources is not
consistent over time – with daily links to media sources ranging between 15 and 0. A closer look
at the data reveals that the largest spikes in media links occur in late August and in mid
November. Although a definitive conclusion about the causes of the increases and decreases in
the use of media links is beyond the scope of this paper, it can be safely concluded that
mainstream media sources do not attract the attention of ordinary political bloggers in the same
way at all times.
Discussion
So what do these findings suggest about the echo chamber hypothesis? According to the
echo chamber hypothesis, the relationship between media coverage on an issue and the number
of blog posts should be positive (r > 0), strong (b closer to 1 than to 0) and tight (a high R²). The
data presented here suggest that the echo chamber hypothesis is accurate for some issues but not
for others and, more importantly, that there is considerable variation in the propensity of
individual political bloggers to act as echo chambers on certain issues. With this conclusion in
mind, what kinds of generalizations might be drawn about the kinds of issues and the kinds of
bloggers that fit the predictions of the echo chamber hypothesis?
22
With respect to the kinds of issues that are likely to exhibit an echo chamber effect, there
appears to be no greater correspondence between media coverage and blog posts on domestic
versus foreign policy issues. Indeed, in the overall sample, significant relationships between
media coverage and blog posts were uncovered on four domestic policy issues – abortion, the
military draft, moral values and stem cell research – and four foreign policy issues – Iraq,
prisoner abuse at Abu Ghraib, the crisis in the Sudan and the hunt for Osama bin Laden. The
distribution of significant relationships was roughly similar even within partisan groups.
Republicans displayed significant relationships on four domestic policy issues – abortion, the
military draft, moral values and stem cell research – and two foreign policy issues – the crisis in
the Sudan and the hunt for Osama bin Laden – while Democrats displayed significant
relationships for three domestic policy issues – same-sex marriage, moral values and stem cell
research – and three foreign policy issues – Iraq, the prisoner abuse scandal at Abu Ghraib and
curtailing access to weapons of mass destruction. Furthermore, in addition to the issues that did
display significant relationships, foreign policy issues were not more likely than domestic policy
issues to display no significant relationship. More precisely, five of the eleven foreign policy
issues in this study (Iran, North Korea, Afghanistan, the war on terrorism and Al Qaeda)
displayed no significant correlation between media coverage and blogs posts and of the eight of
the fourteen domestic policy issues (immigration, affirmative action, welfare, No Child Left
Behind, homeland security, the advertising campaign by the Swift Boat Veterans for Truth, the
forged documents controversy involving Dan Rather and Micheal Moore’s Fahrenheit 9/11)
showed no relationship. Overall, therefore, political bloggers do not appear to be more
responsive to the messages of the mainstream media on foreign policy issues.
Although political bloggers are not more likely to echo the messages of the mainstream
media on foreign policy issues in general, there is one subset of domestic policy issues that
23
political bloggers are particularly unlikely to display an echo chamber effect on: issues relating
to the development of the presidential campaign. To put it simply, the data presented in this
study show no relationship whatsoever between mainstream media coverage and the number of
blog posts on any of the events – such as the advertising campaign by the Swift Boat Veterans
for Truth, the forged documents controversy involving Dan Rather and Micheal Moore’s
blockbuster Fahrenheit 9/11 – that were widely claimed to have a dramatic impact on both
public opinion and on the dynamics campaign strategy. In fact, there was no significant
relationship between media coverage on these issues and political blogs in a single one of the 25
political blogs included in this study. These findings are particularly surprising given that one
recent study by Cornfield et al. (2005) has found that A-list political blogs did respond in
significant ways to media coverage on campaign developments such as the Osama bin Laden
tape, Kerry’s mention of March Cheney and the Sinclair Broadcast Group’s decision to air an
anti-Kerry documentary. Ordinary political bloggers, it appears, seem more likely to echo media
messages when they relate to the substance of politics rather than the developments of political
campaigns.
Making generalizations about the kinds of issues that Democratic and Republican
political bloggers are responsive to is difficult. As suggested above, Democrats and Republicans
are equally responsive to media coverage on foreign policy and domestic policy issues.
Interestingly, however, the specific foreign and domestic policy issues that they respond to are
very different. Democrats, for example, show a significant relationship between media coverage
and blog posts on Iraq, Abu Ghraib, same-sex marriage, weapons of mass destruction, moral
values and stem cell research whereas Republicans show a significant relationship on abortion,
Osama bin Laden, the military draft, the Sudan, moral values and stem cell research. One
possible explanation for these partisan differences is that other, more partisan channels of
24
communication – such as A-list political blogs, messages from the political parties and opinion-
based publications – were reinforcing the mainstream media’s coverage of certain issues and, as
a result, partisan bloggers were more likely to blog on these issues. Unfortunately, the data
presented here cannot be used to assess this hypothesis. Ultimately, content analysis of these
other channels of communication will have to be conducted in order to determine why
Democrats and Republicans differ so dramatically in the issues they are likely to echo.
With respect to the kinds of political blogs that are likely to echo the messages of the
mainstream media, the conclusions are more obvious. To state it simply, political blogs that are
more active (i.e. have a higher average number of posts per day) are more likely to display an
echo chamber effect on all issues than political blogs that are less active. The reasons for this are
probably straightforward: those who blog about political issues more often are probably more
likely to pay care about politics more and, as a result more likely to follow news coverage
closely and those who blog about political issues more often are probably in greater need of
material to blog about and, therefore, more likely to respond to media coverage of political
issues.
Conclusion
I began this paper by arguing that political blogs are best understood as a new form of
political expression. Although this claim may appear somewhat innocuous on its face, I have
argued here that it is crucial to directing our attention towards a number of important questions.
In this paper, I have tried to answer one of these questions – whether the political blogosphere is
an “echo chamber” for mainstream media messages? The data presented here suggest that
ordinary political blogs do sometimes act like echo chambers for mainstream media messages
and sometimes do not. More specifically, this paper has shown that there are significant
differences between the issues that Republicans and Democrats echo and that prolific political
25
bloggers are more likely than unprolific political bloggers to follow media coverage on political
issues.
The findings of this paper suggest a number of directions for future research. First, future
research into the existence of an echo chamber effect in the political blogosphere should extend
the analysis presented here by incorporating more mainstream media sources than I have used
here. To state the matter simply, my measure of mainstream media coverage – the New York
Times – may be a poor measure and, as a result, the results of my analysis could be misleading.
Indeed, if the coverage presented in the New York Times is significantly different from the
coverage of other mainstream media outlets and political bloggers are more likely to turn to these
other mainstream media outlets for their news, the findings presented here may be biased in
favor of or in opposition to the echo chamber hypothesis. In order to avoid this problem, future
research should follow the lead of Cornfield et al (2005) and conduct content analysis on a wide
variety of mainstream media sources – ranging from newspapers to cable news broadcasts.
Second, in addition to including more mainstream media sources in the analysis, future
work should look beyond the mainstream media and investigate the possibility of other kinds of
echo chamber effects. More precisely, the possibility that ordinary political blogs are echo
chambers for the messages of A-list political blogs, partisan newsletters, candidate speeches or
opinion journals needs to be tested alongside the hypothesis that ordinary political bloggers echo
the messages of the mainstream media. Given that some have suggested that political bloggers
are individuals who distrust mainstream media coverage and, therefore, are less likely to echo
mainstream media messages, testing for an echo chamber effect with these alternative channels
of communication is even more important. In other words, political blogs may be an echo
chamber for some sources other than the mainstream media.21
21 Cornfield et al (2005) have taken an initial step in this direction by measuring campaign messages along with mainstream media coverage but their analysis focuses only on A-list political blogs and not on the more numerous ordinary political blogs.
26
Finally, in addition to testing the claim that the ideas discussed by political bloggers are
the same as the issues covered in the mainstream media, future work on political blogs also
needs to analyze the other two dimensions of the echo chamber hypothesis: that the positions
taken by political bloggers on political issues greatly resemble the positions taken by the
blogger’s own party leaders and that the frames and justifications employed by political bloggers
are the same as the frames and justifications of the blogger’s party leaders. This paper’s data
measures only the amount of discussion on certain political issues and not the character of it. As
a result, it is unclear whether media messages are being criticized, supported or juxtaposed with
messages from other sources. Saying that Republicans are more responsive to media messages
on abortion, for example, cannot tell us whether they are responding by repeating the frames and
justifications of their party leaders or by criticizing them. A more detailed content analysis that
takes the issue mention as the unit of analysis rather than the blog post as the unit of analysis
would provide the type of data to explore these other components of the echo chamber
hypothesis.
27
Appendix: Issue Keywords
The issue to be measured is listed in bold. The keywords that counted as a mention of the issue are listed below.
AbortionAbortionPro-choiceRight to chooseRight to LifePro-life
Abu GhraibAbu GhraibPrisoner abuse
Affirmative ActionAffirmative actionReverse discrimination
AfghanistanAfghanistan
Al QaedaAl Qaeda
Bin LadenBin Laden
Dan RatherDan Rather
DraftDraft
Fahrenheit 9/11Fahrenheit 9/11
Gay MarriageGay marriageSame-sex marriageHomosexual marriage
Homeland SecurityHomeland Security
ImmigrationImmigrantsImmigrationIllegal aliens
IranIran
IraqIraqIraqi
Moral ValuesMoral values
No Child Left BehindNo Child Left Behind
28
North KoreaNorth Korea
Patriot ActPatriot Act
Social SecuritySocial Security
Stem CellStem cell
SudanSudan
Swift boat veteransSwift boat veterans
TerrorTerrorTerroristTerrorism
Weapons of Mass DestructionWMDWeapons of Mass Destruction
WelfareWelfare
29
Bibliography
Adamic, Lala and Natalie Glance. 2005. “The Political Blogosphere and the 2004 Presidential Election.” Available at: http://www.blogpulse.com.
Blood, Rebecca. 2000. “Weblogs: A history and perspective.” rebecca's pocket. Retrieved from http://www.rebeccablood.net/essays/weblog_history.html.
Blood, Rebecca. 2002. We've Got Blog: How Weblogs are Changing Our Culture. Cambridge MA: Perseus Publishing.
Bloom, Joel. 2003. “The Blogosphere: How a Once Humble Medium Came to Drive Elite Politics.” Paper presented at the annual meeting of the American Political Science Association Philadelphia Marriott Hotel, Philadelphia, PA.
Business Wire. 2003. “The Blogging Iceburg: of 4.12 Million Weblogs, Most Little Seen and Quickly Abandoned, According to Perseus Survey.” Business Wire. Cambridge, Mass.
Drezner, Daniel and Henry Farrell. 2004. “The Power and Politics of Blogs.” Paper presented at the annual meeting of the American Political Science Association.
Lenhart, Amanda, Horrigan, John B., and Fallows, Deborah. 2004. “Content Creation Online.” PEW Internet & American Life Project. Retrieved Aug. 30, 2004 from http://www.pewinternet.org/pdfs/PIP_Content_Creation_Report.pdf.
McKenna, Laura and Pole.Antoinette. 2004. “Do Blogs Matter? Weblogs in American Politics.” Paper presented at the annual meeting of the American Political Science Association.
Mortensen, Torrill and Jill Walker. 2002. “Blogging Thoughts: Personal Publication as an Online Research Tool,” in Andrew Morrison, ed., Researching ICTs in Context. Oslo: Intermedia.
Schiano, Diane, Bonnie Nardi, Michelle Gumbrecht and Luke Swartz. 2004. “Blogging by the Rest of Us.” Retrieved August 25, 2005 from http://home.comcast.net/~diane.schiano/CHIO4.blog.pdf.
Scott, Esther. 2004. “’Big Media’” Meets the ‘Bloggers’: Coverage of Trent Lott’s Remarks at Strom Thurmond’s Birthday Party.” John F. Kennedy School of Government, Harvard University.
Sunstein, Cass. 2001. Republic.com. Princeton, NJ: Princeton University Press.
Zaller, John. 1992. The Nature and Origins of Mass Opinion. Cambridge: Cambridge University Press.
30
Figure 1 – Links to Mainstream Media Sources
Figure 1: Links to Mainstream Media Sources
0
2
4
6
8
10
12
14
16
7/1/
2004
7/11
/2004
7/21
/2004
7/31
/2004
8/10
/2004
8/20
/2004
8/30
/2004
9/9/
2004
9/19
/2004
9/29
/2004
10/9/
2004
10/19
/200
4
10/29
/200
4
11/8/
2004
11/18
/200
4
11/28
/200
4
31
Table 1 – Final Sample of Political Blogs
Blog Title Address2nd Battalion 94th Artillery http://294fa.blogspot.com/AgnostoLibertarianTechnoGeek http://ddhead.blogspot.com/An etherealgirl's adventure http://etherealgirl.blogspot.comAnti-Climacus http://troester.blogspot.com/Buffalo Pundit http://wny4clark.blogspot.com/Cabin Fever Online http://www.cabinfever.modblog.comCaveat Aranea http://caveataranea.blogspot.comEricRosenfield.com http://ericrosenfield.comGeorge In Denver http://georgeindenver.blogspot.comJustin Blanton http://justinblanton.comMiranda's Window Dressing http://mirandaswindowdressing.blogspot.comNeoconservadroid http://neoconservadroid.blogspot.com/Occasional Blog http://occasionalblog.blogspot.comOdious Woman http://odiouswoman.blogspot.comPatriot Blog http://patriotblog.comRoblog http://www.roblogpolitics.blogspot.com/Sandcastles and Cubicles http://sandcastlesandcubicles.blogspot.comShepherd's Pie, Web Edition http://kenshepherd.blogspot.comStephanie Chu http://www.stephaniechu.com/The Doctor's Log http://masterdev.dyndns.dk/drslogThe Patriette http://thepatriette.com/The Turnspit Daily http://www.turnspit.comThe Weekly Subversive http://weeklysubversive.blogspot.com/The World According to Nick http://schweitn.blogspot.com/Vulgar Boatman http://vulgarboatman.blogspot.com
32
Table 2 – Relationship between media coverage and blog posts in the overall sample
Issue R R² BAbortion .22*** .05 .16**Abu Ghraib .26*** .07 .09***Affirmative Action -.07 .00 .00Afghanistan .00 .00 .00Al Qaeda .12 .01 .03Bin Laden .22*** .05 .18**Dan Rather -.01 .00 .00Draft .21*** .05 .21**Fahrenheit 9/11 -.11 .01 .00Gay Marriage .13 .02 .18*Homeland Security -.08 .01 .00Immigration .03 .00 .00Iran .15 .02 .06*Iraq .25*** .06 .13**Moral Values .44*** .19 .40***No Child Left Behind -.04 .00 .00North Korea -.02 .00 .00Stem Cell .31*** .10 .13***Sudan .21** .04 .06Swift Boat -.09 .01 -.03Terrorism -.07 .00 -.01Welfare -.04 .00 .00WMD .16 .02 .12*Entries for B are unstandardized regression coefficients.*** p < .01 ** p < .05 * p < .10
33
Table 3 – Relationship between media coverage and blog posts by partisanship
Democrats RepublicansIssue r R² B R R² BAbortion .10 .01 .05 .24*** .06 .08**Abu Ghraib .31*** .10 .09*** -.02 .00 .00Affirmative Action -.06 .00 -.02 -.05 .00 -.01Afghanistan .09 .01 .02 .01 .00 .00Al Qaeda .01 .00 0 .14 .02 .03Bin Laden .11 .01 .06 .22*** .05 .10**Dan Rather -.02 .00 .00 .00 .00 .00Draft -.05 .00 -.03 .31*** .10 .24***Fahrenheit 9/11 -.11 .01 .00 -.04 .00 .00Gay Marriage .22*** .05 .23** -.03 .00 -.02Homeland Security -.07 .00 .00 -.04 .00 .00Immigration .02 .00 .00 .01 .00 .00Iran .11 .01 .03 .11 .01 .03Iraq .22*** .05 .08** .10 .01 .03Moral Values .46*** .21 .33*** .18** .03 .08**No Child Left Behind -.02 .00 .00 -.04 .00 .00North Korea -.11 .01 -.01 .06 .00 .01Stem Cell .30*** .09 .09*** .16** .03 .04**Sudan .09 .01 .01 .21*** .04 .04**Swift Boat -.06 .00 .00 -.10 .01 -.02Terrorism -.10 .01 -.02 .01 .00 .00Welfare -.01 .00 .00 -.05 .00 .00WMD .19** .03 .13** .01 .00 .00Entries for B are unstandardized regression coefficients.*** p < .01 ** p < .05 * p < .10
34
Table 4 – Relationship between media coverage and blog posts by individual blog
2nd
Battalio
n
Ac
cord
ing
to N
ick
Ag
no
sto
Lib
erta
rian
An
ti-Clim
acu
s
Bu
ffalo
Pu
nd
it
Cav
eat A
ran
ea
Cab
in F
eve
r On
line
Eric
Ro
sen
field
Eth
erea
l Girl
Geo
rge
in D
en
ver
Jus
tin B
lan
ton
Miran
da's W
ind
ow
Neo
co
ns
ervad
roid
Abortion .11 .15 -.05 .27*** .20** .16** † † † .20** † † .03Abu Ghraib -.02 † † † .13 † † † † † † † .00Affirmative Action -.04 -.04 † † -.04 † † † † † † † †Afghanistan -.14 .05 † -.20** .11 † † † † -.04 † † .10Al Qaeda -.02 † .15 -.02 .01 † † † † -.04 † † .09Bin Laden .35*** -.08 † .12 .06 † † † † -.06 † † -.03Dan Rather -.03 .11 † -.03 -.03 .00 † † † † † † †Draft .24*** -.04 † -.03 .01 † † † -.03 † † † †Fahrenheit 9/11 -.07 .05 † † † † † † † † † † †Gay Marriage † -.03 -.05 -.09 .07 .09 † † † .22*** † † .09Homeland Security -.02 .00 † † -.03 .00 † † † -.02 † -.02 †Immigration -.07 .00 .05 † .05 -.07 † † .00 † † -.03 †Iran .18** -.09 † -.02 -.06 † † † † .19** † -.03 .07Iraq -.02 .05 .11 .20** .20** .14 † .01 .11 .07 † .07 .04Moral Values † .20** -.02 -.02 .43*** .20** † † † .29*** † † .41***No Child Left Behind † † † † -.01 † † † .01 † † † †North Korea .06 .09 † -.11 -.05 .09 † † † .00 † † -.02Patriot Act .01 † † † .07 † † † † .10 † † †Social Security † .11 † † .10 .11 .02 † † .08 † † †Sudan † † † .07 † † † † † † † † †Stem Cell † .08 † † † † † † † .45*** † .12 †Swift Boat -.08 -.02 † † -.05 † † † -.02 † † † †Terrorism .03 -.05 .02 -.10 -.05 -.10 .01 † -.04 -.03 † † -.08Welfare -.03 -.03 -.02 -.01 -.05 † † † † † † † †WMD .02 .04 † -.04 .20** .04 † † † .13 † † -.10*** p < .01 ** p < .05 * p < .10† No posts on this issue.
Table 4 (continued) – Relationship between media coverage and blog posts by individual blog
35
Occ
asio
nal B
log
Od
iou
s W
om
an
Pa
triot B
log
Ro
blo
g
Sa
nd
cas
tles
Sh
ep
herd
's P
ie
Ste
ph
an
ie C
hu
Th
e D
oc
tor's L
og
Th
e P
atriette
Th
e T
urn
sp
it Daily
Vu
lga
r Bo
atm
an
We
ekly S
ub
versiv
e
Abortion .03 † -.01 -.07 .15 .13 † .11 † -.09 .21** †Abu Ghraib † † † † -.04 .04 † .64*** † .07 .06 †Affirmative Action -.04 † † † † † † † † -.04 † †Afghanistan .02 † -.01 † .11 .10 † .01 -.10 .09 -.11 †Al Qaeda † † .00 † .42*** -.07 † † † -.02 -.04 †Bin Laden .00 -.06 -.08 † .00 -.03 † † † .12 .10 †Dan Rather † † .00 .00 .00 -.04 † † † † † †Draft † † .23*** † .23*** -.03 † -.03 † -.05 -.03 †Fahrenheit 9/11 † † -.09 † .08 -.02 † † † -.09 -.10 -.10Gay Marriage .12 † -.07 † .03 .00 † † † .09 .11 †Homeland Security -.04 † -.03 † -.02 -.02 † -.02 † -.03 -.02 †Immigration † † .00 † .04 .09 † † † .02 † †Iran † .04 .09 † .02 .04 † † † .09 -.05 †Iraq -.06 .11 .01 -.02 .06 .13 † -.01 .05 .16** .04 -.12Moral Values † † † † -.02 -.02 † † † -.02 † †No Child Left Behind † † -.02 † -.02 -.02 † † -.02 -.02 † †North Korea † † † † -.02 .08 † † † -.12 -.02 †Patriot Act † † † † .02 † † † † † † †Social Security † † † † .06 .03 † † † † .02 †Sudan † † † † .04 .28*** † † † .09 † †Stem Cell † † .20** † † .03 † † † .07 † †Swift Boat † † -.03 -.02 -.04 -.04 † † -.03 -.03 -.03 †Terrorism -.08 -.04 -.01 .09 .13 -.08 † -.13 -.05 -.01 .00 -.04Welfare † † † .01 -.04 † † -.03 † .07 -.03 †WMD † .14 -.06 † -.11 .09 † † † .06 -.08 .23****** p < .01 ** p < .05 * p < .10† No posts on this issue.
36