2

USING TIME SERIES ANALYSIS TO MEASURE INTERMEDIA AGENDA-SETTING INFLUENCE IN TRADITIONAL MEDIA AND POLITICAL BLOG NETWORKS

The tense relationship between traditional media and the net- worked blogosphere has been a feature of the new mass media land- scape since political blogs arose as a force in the aftermath of the 9/11 U.S. terrorist attacks and with the ensuing wars in Afghanistan and Iraq.1 Traditional media entities are now facing a sea change in the economics of news production and distribution.2 As the printed medi- um continues to lose ground to an online model of production and dis- tribution,3 traditional media entities have been forced to experiment with Web 2.0 technologies while adopting blogging4 and other social media forms5 in an effort to draw active web publics into their news sites.

During the 2008 presidential election, visits to some popular poli- tical blogs increased by triple-digit percentage points over readership of those blogs in the preceding 2007 year.6 Political citizen journalism also continued to mature: citizen blogger and Huffington Post’s OffTheBus contributor Mayhill Fowler broke two of the more sensational campaign stories during the runoff for the 2008 democratic presidential nomina- tion,7 recalling previous success that top political bloggers have had in making public officials accountable.8 Though traditional newsrooms

Sharon Meraz is an assistant professor in the Department of Communication at the University of Illinois, Chicago.

JOURNALISM & MASS COMMUNICATION QUARTERLY176

By Sharon Meraz

J&MC Quarterly Vol. 88, No. 1 Spring 2011 176-194 ©2011 AEJMC

Using time series analysis to gauge intermedia agenda setting in a sam- ple of eighteen U.S. political blogs, two elite traditional news entities, and their eleven political newsroom blogs across three issues in 2007, this study reveals that traditional media were unable to set political blog agendas. Ideologically diverse political blog networks were also able to set traditional media’s online news agenda, and, to a lesser extent, their newsroom blog agenda. Findings point to a dilution of traditional media’s singular agenda-setting influence and to greater interdepen- dence between traditional media and political blogs than suggested by hyperlink studies.

177USING TIME SERIES ANALYSIS TO MEASURE INTERMEDIA AGENDA-SETTING INFLUENCE

were reluctant to quote Fowler as their news source,9 newsrooms from the New York Times to news blog sites such as ABC’s The Note and National Journal’s BlogMeter now regularly access blogs for story leads. These anecdotal reports suggest vibrant informational codependence between political blogs and traditional media outlets; however, little quantitative research has compared how issue emphasis and selection can be used to gauge intermedia agenda-setting influence between these two media entities.

This study works within the methodological tradition of pre- vious studies that have adopted the time series method of Granger causality to conduct agenda-setting analysis.10 Unlike earlier studies, this study measures the number of unique blog posts and news articles written on an issue as an indicator of intermedia issue agenda-setting effects. A previous study11 revealed that political bloggers were equally as likely to link in their blog postings to traditional media as citizen media; however, traditional media were unwilling to link to political blogs. Departing from the use of hyperlink analysis as a marker of source agenda-setting relationships, this study examines how the choice to write blog postings and/or articles about an issue, in contrast to the viewpoints adopted on those issues, can be used as an indicator of inter- media agenda-setting influence. This study compares the quantity of blog postings on three popular public affairs issues in 2007 through a sample of eighteen top independent political blogs (six left-leaning, six right-leaning, and six moderate); two elite traditional media entities; and the latter’s eleven newsroom blogs that center on U.S. political affairs. This study also establishes the relevancy of intermedia agenda- setting theory in explicating flows of influence between both traditional media and political blogs and among political blogs of diverse political ideology.

Intermedia Agenda Setting. Agenda setting remains one of the most enduring and most researched theories in mass communication and political communication.12 The maturation of agenda-setting theory over several decades could explain previous13 findings regarding the leveling off of academic publications that use agenda setting as a theoret- ical device. Since its first appearance in 1972, agenda setting has matured as a theory to include a second-level agenda-setting component (attri- bute agenda setting), a psychological component to explain individual- level agenda-setting effects (need for orientation), an emphasis on how the media’s agenda is shaped, and an explanation for the shared news agenda among different media (intermedia agenda setting). Given its endurance and maturity as a mass communication theory, it is surprising that few researchers directly apply agenda setting, specifically interme- dia agenda setting, to flows of influence that occur between traditional media and the networked political blogosphere as well as among ideo- logically diverse U.S. political blogs.

Intermedia agenda setting seeks to examine how the media’s agenda is set by sources,14 and intermedia agenda setting explains the

Literature Review

flow of influence among media entities.15 In terms of more traditional media, the intermedia agenda-setting influence has been most evident in situations of elite influence, such as elite newsrooms’ influence on less elite mass media entities,16 newspaper influence on television news broadcasts,17 and online newspapers’ influence on wire services.18 Television outlets have also had influence on each other’s agendas. Other forms of influence include political advertising on television and newspaper news19 and advertising by political candidates on tra- ditional media agendas.20 Website agendas of political candidates and traditional mass media agendas have also exhibited strong correla- tions.21

Though research on the intermedia agenda-setting relationship between traditional media and newer, emergent media is limited, most previous research suggests that within the United States, traditional media institutions still have the power to set non-traditional media agendas at the issue level. Traditional media were shown to have influ- ence on the content of the 1996 Republican candidates’ press releases as presented on candidate websites.22 In the 2004 presidential election, tra- ditional media also influenced the campaign blog agendas, as opposed to influence occurring in the reverse direction.23 In the context of blogs, one study did find blog-to-media influence at the issue level; however, the reverse level of influence was stronger.24

Intermedia agenda-setting theory has also been applied to infor- mational influence among non-traditional media entities. One study found significant inter-candidate issue agenda setting during the 2004 U.S. presidential election based on candidate press releases downloaded from the candidate websites.25 Evidence has also been found for shared issue agendas among traditional media platforms and non-traditional portal news outlets, such as Yahoo News and Google News. In terms of the political blogosphere, no studies have utilized intermedia agenda setting to explain flows of influence.26

Previous studies on informational influence and agenda sharing within the political blogosphere have centered on gauging source influ- ence through the analysis of hyperlink connections. Several of these studies have shown that the political blogosphere is stratified along par- tisan lines. Blogs that share political ideology are more prone to link to one another and share similar source agendas.27 These hyperlink studies suggest that partisan affiliation strongly impacts intermedia agenda set- ting through source agendas.

Unlike previous studies on the political blogosphere, this study utilizes the theory of intermedia agenda setting to gauge two levels of influence. First, intermedia agenda setting is used to gauge the influence of the traditional newsroom on political blogs. Second, intermedia agen- da setting is used to assess the flow of influence among political blogs of disparate political ideologies. Departing from previous studies cen- tered in hyperlink analysis, this study examines intermedia agenda set- ting through time series analysis estimated by Granger causality, utiliz- ing the quantity of unique blog postings and articles on select issues as an indicator of intermedia agenda-setting effects.

178 JOURNALISM & MASS COMMUNICATION QUARTERLY

179USING TIME SERIES ANALYSIS TO MEASURE INTERMEDIA AGENDA-SETTING INFLUENCE

Before conducting specific time series tests for intermedia agenda setting, this study will test for evidence of correlated issue agendas as a surface marker of intermedia agenda setting. Given previous findings of ideologically driven hyperlinking practices among partisan political blogs,28 this study advances the following hypothesis in relation to issue selection and emphasis on partisan blogs:

H1: Blogs that share partisan ideologies will have a more correlated issue agenda within their blog posts than that of ideologically disparate blogs.

With the maturation of the political blogosphere, many former journalists are now joining the ranks of elite political bloggers.29 Though many elite political bloggers now gain first-hand access to sources,30most elite traditional media entities still remain in a better structural position to break news and conduct investigative reporting on their online news websites due to their professional affiliation to the newsroom.31 Initiating an intermedia agenda-setting test at the most likely level of traditional media textual influence—online news articles—this study advances the following intermedia agenda-setting hypothesis as it relates to issue selection and emphasis:

H2: Traditional media’s online news articles will be more likely to set the issue agenda of political blog postings than the reverse relationship.

This study also predicts that traditional media’s online news agen- da will have more success in influencing issue selection and emphasis within political blog issue agendas than political blogs within their own blog networks. This study advances the following intermedia agenda- setting hypothesis:

H3: Traditional media’s online news articles will be more likely to set the issue agenda of political blog postings than polit- ical blogs will be at setting their own blog post issue agenda.

Can independent political blogs influence issue selection and emphasis within traditional media outlets? One study found that politi- cal blogs exert influence on traditional media’s issue agenda; however, the greater influence was from traditional media to political blogs.32 This study advances the following research questions:

RQ1: To what extent are political blogs successful at setting the issue agenda in traditional media’s online news articles?

RQ2: To what extent are political blogs successful at set- ting the issue agenda in traditional media’s blog postings?

Hypotheses and Research Questions

180 JOURNALISM & MASS COMMUNICATION QUARTERLY

Due to the novelty of traditional media newsroom blogging, schol- arly attention has yet to assess its agenda-setting impact on other media agendas. Thus, this study advances the following research question:

RQ3: What role do traditional media blogs play in set- ting the agenda of other media entities?

These three hypotheses and three research questions were tested through comparing the frequency of blog postings and news articles written on three select public affairs issues in 2007 in a sample of eight- een elite political blogs across the ideological spectrum, two elite tradi- tional media entities (the New York Times and the Washington Post), and the latter’s eleven newsroom blogs that center on politics.

Past studies reveal that traditional media pay close attention to the content from political blog elites.33 Using quantitative techniques,34 the most popular blog listings from blog aggregators/search engines Technorati, the Truth Laid Bear, and BlogPulse were correlated in an effort to sample elite left-leaning and right-leaning political blogs. This process yielded the following sample of twelve blogs: The Daily Kos (left-leaning), Crooks and Liars (left-leaning), Think Progress (left-leaning), Talking Points Memo (left-leaning), The Huffington Post (left-leaning), FireDogLake (left-leaning), Instapundit (right-leaning), Michelle Malkin (right-leaning), Hot Air (right-leaning), Little Green Footballs (right-lean- ing), Powerline (right-leaning), and Captain’s Quarters (right-leaning). Because of their absence from popular blog listings, it was necessary to utilize a series of triangulation techniques35 to sample popular moderate blogs. This resulted in the selection of Donklephant, The Moderate Voice, The Daily Dish, The Gun Toting Liberal, Central Sanity, and The Van Der Galien Gazette. In total, eighteen blogs were selected for analysis, with six selected across each dominant ideological spectrum.

Previous studies have found that elite newsrooms are agenda set- ters for less elite news media;36 as such, the online versions of the New York Times and the Washington Post were selected to represent the tradi- tional media’s online news agenda. This study also examined the impact of the newsroom’s political blogs, which included the following: The Caucas, The Lede, and the Opinionator for the New York Times, and the political blogs White House Watch, Bench Conference, Achenblog, Early Warning, Think Tank Town, The Fix, On Balance, and OFF/Beat from the Washington Post. Individual media agendas were further aggregated into five broad “network” agendas: the left-leaning blog network, the right-leaning blog network, the moderate blog network, the traditional media network, and the traditional media blog network.

Rather than hand-selecting specific issues, issues for this study were selected based on quantity of discussion generated. The issue-rich period of July 20, 2007, to September 30, 2007, was chosen as a good time period for selecting popular issues, and two primary criteria ultimately guided issue selection. Each issue needed to be bound by a definite start

Method

181USING TIME SERIES ANALYSIS TO MEASURE INTERMEDIA AGENDA-SETTING INFLUENCE

and end point, a factor important in measuring agenda-setting effects; and issues needed to generate notable discussion across all blogs and media. These criteria yielded three issues: the Alberto Gonzales hearings on the NSA wiretapping, the Larry Craig sex scandal, and the Petraeus report and the MoveOn “Betray us” ad on Iraq.

Exact start and end time points for the study of each of the issues were determined by two additional factors: summing the total amount of unique blog posts and articles on each issue through the entire time peri- od for a visible message burst, and ensuring that over 50% of media enti- ties (blogs and traditional media) carried a message about the issue. The start and end time periods in 2007 for the Gonzales NSAwiretapping tes- timony, the Larry Craig scandal, and the Petraeus report and the MoveOn “Betray us” ad on Iraq were July 24 to August 2, August 27 to September 7, and August 15 to September 24, respectively.

The sampling frame was the entire blog and traditional media website, while the unit of analysis was the blog post for blogs (political blogs and traditional media blogs) and the article for traditional media’s online website agenda. For each of the three issues, the advanced search feature on each medium’s website was used to pull all generated content during the time period designated for each issue’s investigation. To ensure reliability regarding which content was counted as related to the select issues,37 inter-coder reliability tests were conducted on a random 10% sample of content for each issue across the media entities. Two coders compared their selection of blog posts and/or online news articles that qualified as posts/articles on each issue. Coders were given specific instructions to select only blog postings or online news articles that had (1) a headline that suggested the posting/article was about the topic, and/or (2) a lead that centrally identified the posting/article with the topic. Inter-coder reliability scores ranged from .94 to .98 using Krippendorff’s alpha, which accounts for chance agreement.38 Upon reaching this acceptable level of agreement, one coder completed the selection of the remaining blog postings and articles that would account for each media entity’s total volume of content for each issue. In total, there were 1,422 unique blog posts (1,262 culled from the eighteen independent blogs in this sample and 160 culled from traditional media blog posts) and 302 traditional media’s online news articles. This count data, aggregated at the network level (left-leaning blog agenda, right- leaning blog agenda, moderate blog agenda, traditional media agenda, traditional media blog agenda) on an issue-by-issue basis (see Table 1), formed the basis for correlating issue agendas (H1) and for testing cau- sation and intermedia agenda setting using Ordinary Least Squares (OLS) regression to estimate Granger causality (H2, H3, RQ1, RQ2, and RQ3).

Time series analysis has long been recognized and utilized as a robust method for determining causation in agenda-setting studies,39 including intermedia agenda-setting effects.40 It is argued that Granger causality can provide a more accurate result than other time series meth- ods, like ARIMA modeling,41 which are more prone to error. Applying Granger causality analysis to this study permitted predictions of each

Media Network Issues Craig Gonzales Iraq

Left 207 123 463 Moderate 73 24 139 Right 49 20 167 TMediaA 46 30 226 TMediaB 39 19 102

Note: Due to space constraints, within Table 1, Left refers to the agenda of left-leaning blogs, Right refers to the agenda of right-leaning blogs, TMediaA refers to the agenda of traditional media’s online news articles, and TMediaB refers to agenda of traditional media’s blog postings.

media network’s agenda based on lagged values of its past agenda and those of other media networks.

Using OLS regression, optimal time lags were tested through the regression of each media network’s agenda against its past agenda until the latter no longer predicted its present agenda. Using STATA to run OLS regression to estimate Granger causality, linear relationships were then developed that regressed each of the five aggregated media net- work agendas based on both their past agenda and the agenda of the other four media networks. For example, traditional media’s agenda- setting influence on the right-leaning blog network would be estimated using both traditional media’s past agenda and the right-leaning blog network’s past agenda as predictors. Running these tests provided t-val- ues of significance based on each medium’s intermedia agenda-setting ability for other media networks.

Given that results of a time series analysis are more reliable with more than 50 time points,42 all 56 usable time points through the three issue periods were pooled to create greater opportunity to deci- pher media and blog influence on each other. At the individual issue level, the only issue that yielded a lengthy enough time period to avoid an overly inflated error term (and a failure to reject the null hypothesis) was the Petraeus Report/MoveOn “Betray Us” Ad issue, which persisted for 40 days. This latter issue provided 38 usable lags for measuring intermedia agenda-setting effects among the five media net- works.

Correlating Issue Agendas Across Media Entities. Before specific hypotheses and research questions were examined, intraclass correla- tion coefficients (ICCs) were calculated, issue-by-issue, on the frequency of postings and article content on each issue in an effort to provide a measure of broad agreement among the five different media networks. ICC is similar to Pearson’s correlation, but useful for non-independent datasets like those in this study.

182 JOURNALISM & MASS COMMUNICATION QUARTERLY

TABLE 1 Unique Blog Posting and Article Volume on Each Issue in Five Media Networks

Results

183USING TIME SERIES ANALYSIS TO MEASURE INTERMEDIA AGENDA-SETTING INFLUENCE

Table 1 provides a summary of issue mentions across the five media networks. Of the three issues, the Petraeus Report/MoveOn issue was the only issue that yielded a significant correlation (r = .415, p < .05) across all five network agendas (left-leaning blogosphere, right-leaning blogosphere, the moderate blogosphere, traditional media’s online news, and traditional media blog). For the Craig issue, the ICC was insignificant (r = .022, p > .05) for broad agreement across the five net- works. Similarly, the ICC was insignificant for the Gonzales issue (r = -.058, p > .05) across all five network agendas.

Pairwise correlations were then run between the networks to determine whether any two networks yielded significant correlation coefficients. As Table 2 highlights, only the Petraeus Report/MoveOn issue yielded significant correlations across all network agendas. Significant correlations on this issue ranged from .272 to .638, and all blog networks showed correlations with the media agenda. Both the left- leaning and the moderate network also had strong correlations with the media blog agenda; however, only the right-leaning blogosphere did not have an agenda that correlated with the media blog agenda.

For the remaining two issues, few significant correlations were derived; yet, a consistent finding was the moderate blogosphere’s strong correlation with traditional media’s blog agenda (a relationship which provides moderate support for H2). For the Craig issue, significant cor- relations were found between the moderate network and the left-leaning blogosphere (r = .456, p < .05), as well as between the moderate blogo- sphere and the media blog agenda (r = .490, p < .05). For the Gonzales issue, the only significant correlation was a correlation between the moderate blogosphere and the media blog agenda (r = .572, p < .05). For all three issues, both the moderate blogosphere and traditional media’s blog agenda shared strong correlations.

ICC findings revealed uncorrelated issue agendas between the left-leaning and right-leaning networks; however, to test H1 fully, it was

TABLE 2 Blog-to-Blog and Blog-to-Media Correlations for the Petraeus Report

and the MoveOn “Betray Us” Ad on the Iraq Issue

Right Left Moderate TMediaA TMediaB

Right .165 .638* .381* .140 Left .308* .474* .308* Moderate .365* .565* TMediaA .272* TMediaB

Note: Due to space constraints, within Table 2, Left refers to the agenda of left-leaning blogs, Right refers to the agenda of right-leaning blogs, TMediaA refers to the agenda of traditional media’s online news articles, and TMediaB refers to agenda of traditional media’s blog postings.

*p < .05

important to correlate the issue agendas of blogs within their ideolo- gical social networks. Such analysis would answer the important ques- tion: Do left-leaning blogs and right-leaning blogs share a similar issue agenda with other blogs within their ideological perspective?

H1 gained consistent support within the left-leaning blogosphere, though the correlation coefficients could be described as weak. The ICC for the agenda of the six left-leaning blogs on the Craig issue was signif- icant (r = .292, p < .05). Regarding the Gonzales issue, the ICC was also significant (r = .225, p < .05) among all six left-leaning blog agendas. Finally, on the Petraeus Report/MoveOn issue, the ICC among the six left-leaning blog agendas proved significant (r = .521, p < .05).

H1 gained only qualified support within the right-leaning blog network, with the strongest support gained for the Petraeus Report/ MoveOn issue. The ICC was highest for this issue (r = .403, p < .05), but barely significant for the Craig issue (r = .131, p = .05); however, the ICC was insignificant for the Gonzales issue (r= -.064, p > .05).

Testing Intermedia Agenda in Traditional Media and Political Blog Networks. These data reveal correlations among agendas among the five different media outlets (left-leaning blogs, right-leaning blogs, moderate blogs, traditional media, and traditional media blogs), but cor- relation coefficients are not evidence of causation. In an effort to deter- mine directional influence, OLS regression was used to estimate Gran- ger causality. These hypotheses also required a test of how effective the media network was at setting its own agenda, which was best predicted by a one-day lag interval (Table 3). Hypotheses were then tested by combining all issues into a pooled issue. Additionally, tests were conduct- ed at the individual issue level using the Petraeus Report/MoveOn issue.

Isolating the most well-known and likely source of traditional media influence, H2 predicted that traditional media’s online news agenda would be more likely to set the issue agenda of political blogs than the reverse relationship. This hypothesis was only partially sup- ported, at both the pooled issue level and at the individual issue level,

184 JOURNALISM & MASS COMMUNICATION QUARTERLY

TABLE 3 One-day Lag Significance Values for Media Network Setting Its Own Agenda

Right Left Moderate TMediaA TMediaB

Right 3.56* . . Left 6.66* . Moderate . . 3.38* . TMediaA 6.39* TMediaB . . 3.69*

Note: Due to space constraints, within Table 3, Left refers to the agenda of left-leaning blogs, Right refers to the agenda of right-leaning blogs, TMediaA refers to the agenda of traditional media’s online news articles, and TMediaB refers to agenda of traditional media’s blog postings.

* p < .05

Granger Test Media Result

Left blog does not Granger cause Right blog 2.83* Moderate blog 1.19 TMediaA 2.62* TMediaB 1.98

Right blog does not Granger cause Left blog 0.53 Moderate blog -0.20 TMediaA 3.18* TMediaB 1.11

Moderate blog does not Granger cause Left blog 1.97 Right blog 2.32* TMediaA 5.02* TMediaB 2.85*

TMediaA does not Granger cause Left blog 1.29 Right blog 2.28* Moderate blog 2.18* TMediaB 3.06*

TMediaB does not Granger cause Left blog 0.47 Right blog 0.58 Moderate blog 0.17 TMediaA 3.04*

Note: Due to space constraints, within Table 4, Left refers to the agenda of left-leaning blogs, Right refers to the agenda of right-leaning blogs, TMediaA refers to the agenda of traditional media’s online news articles, and TMediaB refers to agenda of traditional media’s blog postings. All report- ed results represent t-values between the two media agendas that test Granger causality.

* p < .05

185USING TIME SERIES ANALYSIS TO MEASURE INTERMEDIA AGENDA-SETTING INFLUENCE

because of traditional media’s inability to set the left-leaning blog agen- da. Table 4 presents the results at the pooled issue level, while Table 5 presents these results at the individual issue level. Using the syntax of Granger causality, noting that the words “agenda setting“ are substitut- ed with the word “cause,“43 traditional media online news sites did Granger cause the right-leaning blog agenda through the pooled issue level (t[2, 54] = 2.28, p < .05) and the individual issue level (t[2, 36] = 3.15, p < .05), the moderate blog agenda through the pooled issue level (t[2, 54] = 2.18, p < .05) and the individual issue level (t[2, 36] = 2.05, p < .05) , and its own media blog agenda through the pooled issue level (t[2, 54] = 3.06, p < .05), and the individual issue level (t[2, 36] = 3.64, p < .05). But, tradi- tional media online news sites did not Granger cause the left-leaning blog agenda at either the pooled issue level (t[2, 54] = 1.29, p > .05) or the individual issue level (t[2, 36] = -1.18, p > .05).

H3 predicted that political blogs would not be successful at setting one another’s agenda; however, this was also unsupported. Through the

TABLE 4 Pooled Issue Data Testing Traditional Media to Political Blog Agenda Setting

186

pooled issue level and the individual issue level, political blogs did Granger set each other’s agenda. As Tables 4 and 5 show, the left-lean- ing blog network did Granger cause the agenda of the right-leaning blo- gosphere network at the pooled issue level (t[2, 54] = 2.83, p < .05) and the individual issue level (t[2, 36] = 4.12, p < .05). Right-leaning blogs had less agenda-setting power on other blogs, and in both scenarios (pooled and individual issue level), this network was unable to set the agenda of other political blogs. An interesting scenario of the moderate blog’s power is shown by its agenda-setting ability: the moderate blog network did Granger cause the issue agenda of the right-leaning blogos- phere at the pooled issue level (t[2, 54] = 2.32, p < .05) and the individ- ual issue level (t[2, 36] = 2.24, p < .05). However, the moderate blog net- work did not Granger cause the left-leaning blog network’s agenda at both the pooled issue level and the individual issue level.

Were the political blog networks able to set traditional media’s online news issue agenda (blog and traditional news websites)? In

JOURNALISM & MASS COMMUNICATION QUARTERLY

TABLE 5 Individual Issue Data Testing Traditional Media to Political Blog Agenda Setting

Granger Test Media Result

Left blog does not Granger cause Right blog 4.12* Moderate blog 2.13* TMediaA 3.75* TMediaB 3.36*

Right blog does not Granger cause Left blog 0.10 Moderate blog 0.36 TMediaA 2.38* TMediaB 1.34

Moderate blog does not Granger cause Left blog 0.74 Right blog 2.24* TMediaA 4.22* TMediaB 2.24*

TMediaA does not Granger cause Left blog 0.83 Right blog 3.15* Moderate blog 2.05* TMediaB 3.64*

TMediaB does not Granger cause Left blog -1.18 Right blog 0.22 Moderate blog -0.05 TMediaA 1.76

Note: Due to space constraints, within Table 5, Left refers to the agenda of left-leaning blogs, Right refers to the agenda of right-leaning blogs, TMediaA refers to the agenda of traditional media’s online news articles, and TMediaB refers to agenda of traditional media’s blog postings. All report- ed results represent t-values between the two media agendas that test for Granger causality.

*p < .05

answer to RQ1 and RQ2, Tables 4 and 5 also provide solid evidence that all blog networks had success at setting traditional media’s online news agenda (as opposed to traditional media’s blog agenda) at the pooled issue level and the individual issue level. The left-leaning blogosphere was successful at setting traditional media’s agenda through the pooled issue level (t[2, 54] = 2.62, p < .05) and the individual issue level (t[2, 36] = 3.75, p < .05). Similarly, the right-leaning blogosphere wielded influ- ence over traditional media’s agenda through the pooled issue level (t[2, 54] = 3.18, p < .05) and the individual issue level (t[2, 36] = 2.38, p < .05). Unlike the left-leaning or the right-leaning blogosphere, the moderate blogosphere influenced the issue agendas of traditional media’s online news site through both the pooled issue level (t[2, 54] = 5.02, p < .05) and the individual issue level (t[2, 36] = 4.22, p < .05), as well as traditional media’s blog agenda through both the pooled issue level (t[2, 36] = 2.85, p < .05) and the individual issue level (t[2, 36] = 2.24, p < .05).

RQ3 probed the effectiveness of the traditional media blog as an agenda setter for other blog media agendas. The results reveal that tradi- tional media blogs are less of an influencer and are more influenced by their online news sites and by political blog networks. The left-leaning (t[2, 36] = 3.36, p < .05) and the moderate blogosphere (t[2, 36] = 2.24, p < .05) did Granger cause the traditional media’s blog agenda on the indi- vidual issue level, while the moderate blogosphere also leveraged agen- da-setting power over the media blog at the pooled issue level (t[2, 54] = 2.85, p < .05). Through both the pooled issue level (t[2, 54] = 3.06, p < .05) and the individual issue level (t[2, 36] = 3.64, p < .05), traditional media’s online news agenda did Granger cause their internal media blog agenda.

Previous studies have utilized hyperlink analysis to gauge source influence between political blogs and traditional mass media entities.44 However, source influence based solely on hyperlink analysis provides only a limited, often shorthand measure of agenda-setting influence. Unlike those studies, this study utilized issue selection and emphasis as a yardstick for gauging the intermedia agenda-setting influence among diverse media entities. Unlike previous studies, this study also sought to establish the significance of intermedia agenda-setting theory in explain- ing informational influence between traditional media and the political blogosphere, as well as within the U.S. political blogosphere.

In confirming the findings of previous hyperlinking studies45 that point to a dilution of traditional media’s singular agenda-setting power over all web publics, this study also affirms the growing influence of the progressive political blogosphere in setting other media agendas, while resisting traditional media agenda setting. Progressive blogs now resem- ble traditional media in their embrace of team blogging, investigative journalism, and editorial workflow practices.46 Blogs like the Huffington Post are drawing larger audiences than the Washington Post, the Los Angeles Times, and theWall Street Journal.47 These political blogs also gain energy from the endless cycle of extended U.S. political campaigns, sug- gesting that the popularity of this alternative news stream can further

187USING TIME SERIES ANALYSIS TO MEASURE INTERMEDIA AGENDA-SETTING INFLUENCE

Conclusion

188

dilute the influence of traditional media during heightened political periods.

Though previous hyperlink analysis studies found little evidence of extensive linking on the part of traditional media to political blogs,48 this study’s findings revealed that all blog networks (left-leaning, right- leaning, and moderate) exerted moderate influence on traditional media’s agenda. Moderate support was not surprising, given this study’s basis in issue selection, as opposed to viewpoints. The disjunc- ture in this study’s findings from hyperlink studies could suggest a limitation of using hyperlinks as a surrogate for the media agenda. Traditional media may still be wary of linking to sites, preferring to maintain their walled gardens. Hyperlinking is also the currency for attention and relevance in the networked web environment,49 and since more hyperlinks from elite actors can drive up search engine visibility and web traffic, traditional media may be wary of contributing to their own loss of an audience by ceding ground to this competitive, growing news platform. Drawing from text is less visible than linking to sites.

This study’s moderate findings of mutual intermedia agenda-set- ting influence between traditional media and most blog networks might also be based on the choice of issues studied.50 As has been shown in such past events as the Trent Lott scandal and the Dan Rather resigna- tion, political bloggers and traditional media often display mutual co- dependence in investigating a top political story.51 During heightened political events, it is common for both traditional media and indepen- dent political blogs to utilize each other as sources. Given this study’s limited focus on a small sample of popular, hot-button issues and elite blogs, future studies should examine more diverse events and less elite political blogs to see if blog reputation or issue characteristics might predict routine or limited dependence of traditional media on political blog agendas.

This study is unique in its application of intermedia agenda-set- ting theory to the transference of partisan issue agendas among political blogs. Similar to hyperlinking studies which reveal shared source agen- das among ideologically similar political blogs, blogs that share partisan ideologies are more prone to share an issue agenda. Future studies should explicate how the process of intermedia agenda setting within these partisan platforms is enhanced by social influence theories, like social network science, the two-step flow theory, and homophily. Web 2.0 technologies have enabled social influence to be triggered through the power of influentials or trendsetters who act as hubs for informa- tion. Future studies can explore how social influence theories work to complement and predict the intermedia agenda-setting process within networked environments like the U.S. political blogosphere.

The coherence in partisan blog issue agendas is a significant find- ing that can re-awaken fears of selective attention in the twenty-first century media climate. It is possible that web publics will continue to self-select into these partisan communities because it is now easier.52 Future studies should examine the extent to which these partisan poli- tical blogs and traditional media platforms converge on the most impor-

JOURNALISM & MASS COMMUNICATION QUARTERLY

189

tant news of the day. These future studies are important, given the rising popularity of these partisan political blogs as alternatives to traditional news media platforms, particularly during heightened political events like U.S. presidential elections.

NOTES

1. Sharon Meraz, “The Blogosphere’s Gender Gap: Differences in Visibility, Popularity, andAuthority,” inWomen, Men, and News, ed. Paula Poindexter, Sharon Meraz, and Amy Schmitz Weiss (NY: Routledge, 2008), 129-51; Tim Cavanaugh, “Let Slip the Blogs of War,” in We’ve Got Blog: How Weblogs Are Changing Our Culture, ed. John Rodzvilla (Cambridge, MA: Perseus Publishing, 2002); Lakshmi Chaudhry, “Can Blogs Revolutionize Progressive Politics,” Salon, February 6, 2006, avail- able from http://www.inthesetimes.com/article/2485/can_blogs_ revolutionize_progressive_politics/.

2. Several books treat the changing dynamics of media economics in the twenty-first century. For example, see Chris Anderson, The Long Tail: Why the Future of Business Is Selling Less of More (NY: Hyperion, 2006); Yochai Benkler, The Wealth of Networks: How Social Production Transforms Markets and Freedom (New Haven, CT: Yale University Press, 2007); Axel Bruns, Blogs, Wikipedia, Second Life and Beyond: From Production to Produsage (NY: Peter Lang, 2008); Clay Shirky, Here Comes Everybody: The Power of Organizing Without Organizations (NY: Penguin Group, 2009).

3. See, “Internet Most Popular Information Source: Poll,” Reuters, June 17, 2009, available from http://www.reuters.com/article/newsOne /idUSTRE55G4XA20090617.

4. Meraz, “The Blogosphere’s Gender Gap: Differences in Visibility, Popularity, and Authority.”

5. Many newsrooms have begun transitioning from blogging appli- cations to microblogging platforms, like Twitter, and social networking platforms, such as Facebook, in an effort to reach out to the younger audience, as well as more active web publics. As of 2010, few academic articles have been published on the usage and utility of social media in the newsroom.

6. See “Huffington Post and Politico LeadWave of Explosive Growth at Independent Political Blogs and News Sites This Election Season,” ComScore, October 22, 2008, available from http://www.comscore.com/ Press_Events/Press_Releases/2008/10/Huffington_Post_and_Politico_ Lead_Political_Blogs; Lee Rainie andAaron Smith, “The Internet and the 2008 Election,” Pew Internet andAmerican Life Project, 2008 [cited 2010]; available from http://www.pewinternet.org/PPF/r/252/report_ display.asp.

7. Mayhill Fowler, “Bill Clinton: Purdum a ‘Sleazy‘ ‘Slimey‘ ‘Scum- bag,‘” The Huffington Post, June 2, 2008, available from http:// www.huffingtonpost.com/mayhill-fowler/bill-clinton-purdhum-a- sl_b_104771.html; Mayhill Fowler, “Obama: No Surprise that Hard- Pressed Pennsylvanians Turn Bitter,” The Huffington Post, April 11, 2008,

USING TIME SERIES ANALYSIS TO MEASURE INTERMEDIA AGENDA-SETTING INFLUENCE

190 JOURNALISM & MASS COMMUNICATION QUARTERLY

available from http://www.huffingtonpost.com/mayhill-fowler/ obama-no-surprise-that-ha_b_96188.html.

8. For examination of the role of political bloggers in the 2002 resig- nation of Trent Lott as Senate Republican leader, see Thomas Edsell and Brian Faler, “Lott Remarks on Thurmond Echoed 1980 Words,” Wash- ington Post, December 11, 2002, available from http://www.washington post.com/ac2/wp-dyn/A37288-2002Dec10; John Mercurio, “Lott Apologizes for Thurmond Comment,” CNN, December 10, 2002, avail- able from http://archives.cnn.com/2002/ALLPOLITICS/12/09/ lott.comment/. For the role that Josh Micah Marshall played in bringing about the 2007 resignation of Alberto Gonzales, see Noam Cohen, “Blogger, Sans Pajamas, Rakes Muck and a Prize,” New York Times, February 25, 2008, available from http://www.nytimes.com/2008/02/ 25/business/media/25marshall.html.

9. Jay Rosen, “The Uncharted: From Off the Bus to Meet the Press,” The Huffington Post, April 14, 2008, available from http://www.huffing tonpost.com/jay-rosen/the-uncharted-from-off-th_b_96575.html. 10. One of the more comprehensive overviews of the usage of

Granger causality in agenda-setting studies is detailed by Stuart Soroka in his book, Agenda Setting Dynamics in Canada (Vancouver, BC: UBC Press, 2002). The most notable studies of the usage of Granger causality for measuring agenda-setting effects include the following: Hans-Bernd Brosius and Hans Mathias Kepplinger, “The Agenda Setting Function of Television News: Static and Dynamic Views,” Communication Research 17 (April 1990): 188-211; Kim Smith, “Newspaper Coverage and Public Concern about Community Issues: A Time Series Analysis,” Journalism Monographs 101 (August 1987): 1-32; William J. Gonzenbach, The Media, The President, and Public Opinion: A Longitudinal Analysis of the Drug Issue, 1984-1991 (Mahwah, NJ: Lawrence Erlbaum Associates, 1996). Though intermedia agenda setting is not the primary focus of the following paper, the authors provide evidence of intermedia agenda-setting effects in the influence of the newspaper agenda on TV: Stefaan Walgrave, Stuart Soroka, and Michiel Nuytemans, “The Mass Media’s Political Agenda-Setting Power: A Longitudinal Analysis of Media, Parliament, and Government in Belgium (1993 to 2000),” Comparative Political Studies 41 (June 2008): 814-36. 11. Sharon Meraz, “Is There an Elite Hold? Traditional Media to

Social Media Agenda Setting Influence in Blog Networks,” Journal of Computer Mediated Communication 14 (April 2009): 682-707. 12. See Jennings Bryant and Dorina Miron, “Theory and Research in

Mass Communication,” Journal of Communication 54 (December 2004): 662-704. 13. David Weaver, “Thoughts on Agenda Setting, Framing, and

Priming,” Journal of Communication 57 (March 2007): 142-47. 14. Maxwell McCombs, Setting the Agenda: The Mass Media and Public

Opinion (Cambridge, UK: Polity, 2004). 15. McCombs, Setting the Agenda, 114. 16. Stephen Reese and Lucig Danielian, “InterMedia Influence and

the Drug Issue: Converging on Cocaine,” in Communication Campaigns

191USING TIME SERIES ANALYSIS TO MEASURE INTERMEDIA AGENDA-SETTING INFLUENCE

About Drugs: Government, Media and the Public, ed. Pamela Shoemaker (Hillsdale, NJ: Lawrence Erlbaum, 1989): 29-45. 17. Esteban Lopez-Escobar, Juan Pablo Llamas, Maxwell McCombs,

and Federico Rey Lennon, “Two Levels of Agenda Setting among Advertising and News in the 1995 Spanish General Elections,” Political Communication 15 (March 1998): 225-38; Marilyn Roberts and Maxwell McCombs, “Agenda Setting and Political Advertising: Origins of the News Media,” Political Communication 11 (July-September 1994): 249-62. 18. Jeongsub Lim, “A Cross-Lagged Analysis of Agenda Setting

Among Online News Media,” Journalism & Mass Communication Quarterly 83 (summer 2006): 298-312. 19. Lopez-Escobar et al., “Two Levels of Agenda Setting among

Advertising and News in the 1995 Spanish Elections.” 20. Thomas P. Boyle, “Intermedia Agenda Setting in the 1996

Presidential Primaries,” Journalism & Mass Communication Quarterly 78 (spring 2001): 26-44; Roberts and McCombs, “Agenda Setting and Poli- tical Advertising.” 21. John Tedesco, “Issue and Strategy Agenda Setting in the 2004

Presidential Election: Exploring the Candidate-Journalist Relationship,” Journalism Studies 6 (May 2005): 187-201; Gyotae Ku, Lynda Lee Kaid, and Michael Pfau, “The Impact of Web Site Campaigning on Traditional News Media and Public Information Processing,” Journalism & Mass Communication Quarterly 80 (autumn 2003): 528-47. 22. McCombs, Setting the Agenda, 106. 23. Kaye D. Sweetser, Guy J. Golan, and Wayne Wanta, “Intermedia

Agenda Setting in Television Advertising and Blogs During the 2004 Election,”Mass Communication and Society 11 (spring 2008): 197-216. 24. Michael Cornfield, Jonathan Carson, Alison Kalis, and Emily

Simon, “Buzz, Blogs, and Beyond: The Internet and National Discourse in the Fall of 2004,” Pew Internet and American Life Project, 2005 [cited 2010]); available from http://www.pewtrusts.org/uploadedFiles/ wwwpewtrustsorg/News/Press_Releases/Society_and_the_Internet/ PIP_Blogs_051605.pdf. 25. Tedesco, “Issue and Strategy Agenda Setting in the 2004

Presidential Election,” 187. 26. Jason Yu and Debashis Aikat, “News on the Web: Agenda Setting

of Online News in News Web Sites of Major Newspaper, Television, and Online News” (paper presented at the annual meeting of the Inter- national Communication Association, New York, 2009). 27. Lada Adamic and Natalie Glance, “The Political Blogosphere and

the 2004 U.S. Election: Divided They Blog” (2005 [cited 2006]), available from http://www.blogpulse.com/papers/2005/AdamicGlanceBlog WWW.pdf; SharonMeraz, “Event Blogging the 2004 Conventions: Media Bloggers, Non-Media Bloggers, and Their Network Connections” (paper presented at the annual meeting ofAEJMC, SanAntonio, TX, 2004); Mark Tremayne, Nan Zheng, Jae Kook Lee, and Jaekwan Jeong, “Issue Publics on the Web: Applying Network Theory to the War Blogosphere,” Journal of Computer Mediated Communication 12 (October 2006): 290-310. 28. Adamic and Glance, “The Political Blogosphere and the 2004 U.S.

Election,” 8; Meraz, “Event Blogging the 2004 Conventions”; Tremayne et al., “Issue Publics on the Web,” 14. 29. Most of the most popular political blogs are written by former

journalists. Blogs like Talking Points Memo, Politico, and the Huffington Postwere all started by former journalists or continue to employ journal- ists that depart the traditional media industry. The maturation and growth of the left-leaning blogosphere is well-documented by Matt Stoller, “What Is This New Movement, TPM Café, January 15, 2007, available from http://www.tpmcafe.com/blog/coffeehouse/2007/ jan/15/what_is_this_new_movement. 30. Fowler, “Bill Clinton”; Fowler, “Obama.” 31. Walter Lippman, Public Opinion (New York: Harcourt Brace and

Company, 1922); Maxwell McCombs and Donald Shaw, “The Agenda- Setting Function of Mass Media,” Public Opinion Quarterly 36 (summer 1972): 176-87. 32. Cornfield et al., “Buzz, Blogs, and Beyond: The Internet and

National Discourse in the Fall of 2004,” 17. 33. Meraz, “Is There an Elite Hold?”; Cornfield et al., “Buzz, Blogs,

and Beyond”; Daniel Drezner and Henry Farrell, “The Power and the Politics of Blogs,” Public Choice 134 (January 2008): 15-30; Cameron Marlow, “Audience Structure in the Weblog Community” (paper pre- sented at the annual meeting of AEJMC, New Orleans, 2004). 34. Though many of these blog aggregators use their own individual

algorithms for determining popularity and relevance, and though the actual position of the political blog may differ in popularity as it relates to other genre-related blogs in technology and entertainment, all of these lists correlated on the most popular political blogs when this sam- ple was developed in 2007. These twelve popular partisan blogs were ranked as more popular in a relative sense than other political blogs present in these top 100 blog lists. 35. The selection of moderate blogs involved a more complicated

methodology, since moderate blogs rarely make it to top 100 blog lists due to lower audience readership. As such, popular moderate blogs were identified through the following techniques: (a) presence in the three aforementioned aggregators (Technorati, Truth Laid Bear, and BlogPulse); (b) popularity in the “Moderate Blog Advertising Network,“ a network listing the most popular moderate blogs and the dollar amount to advertise on each blog present in the network; (c) nomination in the WeblogAwards, a popular blog contest running since 2003, which takes net nominations for the most popular blogs, including those writ- ing about politics; and (d) frequency of appearance in blogroll listings within both elite partisan popular blogs and within the blogrolls for other moderate blogs. The use of all of these techniques made it possi- ble to select a sample of popular moderate blogs based on frequency of appearance in all of these indicators. 36. Reese and Danielian, “InterMedia Influence and the Drug Issue:

Converging on Cocaine,” 6. 37. Conducting intercoder reliability studies for quantitative studies

are essential, even for data that is coded in nominal content categories.

192 JOURNALISM & MASS COMMUNICATION QUARTERLY

193USING TIME SERIES ANALYSIS TO MEASURE INTERMEDIA AGENDA-SETTING INFLUENCE

See Matthew Lombard, Jennifer Snyder-Dutch, and Cheryl Campanella Bracken, “Content Analysis in Mass Communication: Assessment and Reporting of Intercoder Reliability,” Human Communication Research 28 (October 2002): 587-604; Steven Lacy and Daniel Riffe, “Sampling Error and Selecting Intercoder Reliability Samples for Nominal Content Categories,” Journalism & Mass Communication Quarterly 73 (winter 1996): 963-73; Richard H. Kolbe and Melissa S Burnett, “Content Analysis Research: An Examination of Applications with Directives for Improving Research Reliability and Objectivity,” Journal of Consumer Research 18 (September 1991): 243-50. 38. See Klaus Krippendorff, Content Analysis: An Introduction to Its

Methodology (Thousand Oaks, CA: Sage Publications, 2004). 39. The majority of time series analysis utilizing Granger causality

has been done at the issue agenda-setting level, though the logic of Granger causality is highly appropriate for testing intermedia agenda setting (see citation 40). Aforementioned issue agenda-setting studies using Granger causality include the following: Brosius and Kepplinger, “The Agenda Setting Function of Television News: Static and Dynamic”; Gonzenbach, The Media, The President, and Public Opinion; Smith, “Newspaper Coverage and Public Concern about Community Issues”; and Soroka, Agenda Setting Dynamics in Canada. 40. Though it was not the original intent, one of the few studies to

reveal intermedia agenda-setting effects using Granger causality is one conducted by Walgrave, Soroka, and Nuytemans, “The Mass Media’s Political Agenda-Setting Power,” 12. 41. See John R. Freeman, “Granger Causality and the Time Series

Analysis of Political Relationships,” American Journal of Political Science 27 (May 1983): 327-58. 42. George E.P. Box and Gwilym M. Jenkins, Time Series Analysis:

Forecasting and Control (San Francisco: Holden-Day Publications, 1976). 43. In utilizing Granger causality analysis to measure agenda-setting

effects, it is common to substitute the words “agenda setting” for cause. In such situations, cause is a word that denotes agenda-setting effects. Upon running Granger causality tests, with the outcome of significant F- tests or t-tests, the independent variable in the equation is said to “Granger cause” the other. 44. Meraz, “Is There an Elite Hold?”; Marcus Messner and Marcia

Watson Distaso, “The Source Cycle: How Traditional Media and Weblogs Use Each Other as Sources,” Journalism Studies 9 (June 2008): 447-63. 45. Adamic and Glance, “The Political Blogosphere and the 2004 U.S.

Election”; Meraz, “Is There an Elite Hold?”; Meraz, “Event Blogging the 2004 Conventions“; Tremayne et al., “Issue Publics on the Web.” 46. Chris Bowers and Matt Stoller, “ The Emergence of the Progres-

sive Blogosphere: ANew Force in American Politics,” New Politics Insti- tute,August 10, 2005, available from http://www.newpolitics.net/node /87?full_report=1; Stoller, “What Is This New Movement”; Matt Stoller, “Moving Away from the 1960s Left,” MyDD.com, December 31, 2006, available from http://www.mydd.com/story/2006/12/31/14534/097.

47. Amanda Ernst, “Huffington Post Defies Expectations, Reaches New Heights Post Election,” FishBowlNY, June 2, 2009, available from http://www.mediabistro.com/fishbowlny/new_media/huffington_ posts_traffic_more_than_doubles_year_over_year_149222.asp. 48. Meraz, “Is There an Elite Hold?” 49. The role of links as the currency for search engine visibility is

documented in several articles and studies. Some pivotal articles include the following: Pan Bing, Helene Hembrooke, Thorsten Joachims, Lori Lorigo, Geri Gary, and Laura Granka, “In Google We Trust: Users’ Deci- sions on Rank, Position, and Relevance,” Journal of Computer Mediated Communication 12 (April 2007): 801-23; Jon Kleinberg, “Authoritative Sources in a Hyperlinked Environment,” Journal of the ACM 46 (September 1999): 604-32; Larry Page, Sergey Brin, Rajeev Motwani, and Terry Winograd, “The PageRank Citation Ranking: Bringing Order to the Web,” (1999 [cited 2010]), available from http://ilpubs.stanford.edu:8090 /422/. 50. See Meraz, “Is There an Elite Hold?” for a full review of the litera-

ture on issue characteristics and its effects on agenda setting. Also see David Weaver, Doris Graber, Maxwell McCombs, and Cham Eyal, Media Agenda-Setting in a Presidential Election: Issues, Images, and Interests (New York, NJ: Praeger, 1981). 51. Meraz, “The Blogosphere’s Gender Gap,” 139. 52. Eric Lawrence, John Sides, and Henry Farrell, “Self Segregation or

Deliberation? Blog Readership Participation and Polarization in American Politics,“ Perspectives in Politics 8 (March 2010): 141-57.

194 JOURNALISM & MASS COMMUNICATION QUARTERLY