The Power of Message Networks: A Big-Data Analysis of the Network Agenda Setting Model and Issue Ownership

The network agenda setting (NAS) model asserts that news media not only tell us what to think and how to think about it but also construct network agendas to determine how we associate different messages (Guo, 2013; Guo & McCombs, 2011a, 2011b). Empirical evidence has been found to support this new media effects model in areas such as political communication, public relations, and public opinion research (e.g., Kiousis et al., 2014; Vargo, Guo, McCombs, & Shaw, 2014; Vu, Guo, & McCombs, 2014). As these studies demonstrate, the NAS model contributes to the discussion of new media and communication theories by enabling a more nuanced representation of information networks and by offering an approach to better measure the flow of such networks between different entities (e.g., news media, the public, political organization). Maxwell McCombs (2014), one of the founders of agenda-setting, coined the model as ‘‘the third level’’ of the theory in his recent book Setting the Agenda (p. 55).

We argue that, in addition to its practical applications, the NAS model can also enrich the understanding of other communication and social science theories. This present study showcases its potential contribution to the theory of ‘‘issue ownership,’’ which indicates that voters tend to associate certain issues with certain political parties or candidates (Budge & Farlie, 1983; Petrocik, 1996). Based on the NAS model and issue ownership, we propose a new concept, issue ownership network in this article. It suggests that news media and political campaigns can determine the public’s identification of a political party or a candidate with not just individual issues but also entire networks of issues. In such issue networks, some issues may serve as bridges to link the given party or candidate to other issues. The new theoretical perspective expands the boundary of both the NAS model and the theory of issue ownership. It also provides insights to more effectively construct campaign messages during political elections.

The purpose of this article is twofold. It first seeks to explicate the new concept of issue ownership network. Second, the article aims to empirically test the concept in the new media environment. In light of the increasing number of media platforms, scholars have questioned the applicability of traditional media effects models such as agenda-setting in this complex, fragmented media landscape. Although some studies suggest that traditional news media still lead the public opinion (e.g., Vargo et al., 2014; Weeks & 558 GUO AND VARGO Downloaded by [162.253.25.113] at 08:44 11 September 2015 Southwell, 2010), others found a two-way relationship between the media and public agenda (e.g., Neuman, Guggenheim, Jang, & Bae, 2014; Ragas, Tran, & Martin, 2014). This present study is ambitious in that it doesn’t just ask whether media such as newspapers and television set the public agenda, it explores whether they can do so in ways more complicated through constructing issue networks. Also, the study compares the agenda-setting impact of news media with that of political campaigns.

Methodologically, the study analyzes a large data set of news, campaign and public opinion messages on Twitter against the backdrop of the 2012 U.S. presidential election. Unlike traditional surveys, tweets from ordinary users reflect their unsolicited thoughts about a variety of topics, including their perceptions of political candidates and issues related to them. Big data analytics and semantic network analysis were used to investigate millions of tweets. Conclusions and implications of the new theoretical perspectives are discussed.

 In this article, we present an original, empirical study to illustrate the effectiveness of an emerging media effects model, the NAS model, and demonstrate how the model can enrich other communication and social science theories such as issue ownership through its unique concepts. News media and political campaigns affected the public’s identification of a given candidate with unique issues. They also determined how citizens associated a candidate with a network of issues. This ‘‘issue ownership network’’ effect is illustrated here for the first time. Focusing on the association and interplay between different stakeholders, messages, ideas, and concepts, issue ownership network, and the NAS model, in general, allow researchers to investigate more contextual information embedded in the communicating text. As a result, we can discover more nuanced communication effects. The use of big data analytics and semantic network analysis enables the visualization of message networks on the Twitter-sphere. We can see how the public, the media, and campaigns linked issues together. In a way, this article illustrates different cognitive networks of the election.

In particular, the results shed light on media effects in the context of political elections. Network theorists suggest that when two issues are associated, network phenomena such as bridging can occur. This can result in a transfer of perceived expertise when the competent dimension of the originating issue is positive. The results from this analysis might suggest for instance that public opinion of Obama’s competency in dealing with the issue of taxes—a traditionally Republican issue (e.g., Petrocik et al., 2003)—may have increased because (a) the news media linked the issue with TABLE 6 Top Issue Dyads Perceived by the Liberals (the Competent Level) Preelection Period (August 1–31) Election Period (October 23–November 6) Obama Romney Energy, Jobs Foreign affair, Jobs Economy, Fedbudget Foreign affair, Jobs, Education Foreign affair, Healthcare Economy, Jobs Foreign affair, Energy Economy, Jobs Fedbudget, Education Energy, Jobs, Education Economy, Healthcare Fedbudget, Healthcare Tax, Fedbudget, Foreign affairs Economy, Education Abortion, Healthcare Jobs, Middle-class Foreign affair, Education Note. Fedbudget = federal budget deficit. THE POWER OF MESSAGE NETWORKS 573 Downloaded by [162.253.25.113] at 08:44 11 September 2015 energy in their coverage of Obama, and (b) Obama had a better reputation in dealing with energy issue than did Romney (Koch, 2012). With the substantial support given to issue ownership networks, this article provides, further analysis may tie these results to voting outcomes to further explicate such a ‘‘bridging’’ effect.

Here we also incorporate the distinction of affective sentiment to the NAS model. Previous NAS model studies suggest that news media can tell us how to associate different elements, but none of these studies distinguish between positive, neutral, or negative associations. Borrowing the concept of ‘‘competent’’ issue ownership (Walgrave et al., 2012), the study demonstrates that the positive association between a given candidate’s network of issues and between issues themselves can also be transferred from media and campaign messages to the public’s mind. Thus, this study adds an affective level of analysis to the NAS model. This effect, although less than nonaffective salience transfer, was still significant.

Results also showed that the traditional news media was a stronger predictor of the public’s perception of issue ownership networks when compared to political campaign messages. Indeed, social networking sites and other emerging media platforms have provided unprecedented opportunities for individuals or organizations to broadcast their messages to a large mass audience. Their effects on public opinion are nevertheless limited. The traditional news media still represent the mediated center that stages social reality (Couldry & Curran, 2003). Although the boundary between mass communication and interpersonal communication continues to blur in this digital age, the symbolic power of traditional news media remains. Simply put, news media still set the public agenda, and do so in ways more complicated than previously thought. They construct message networks. To further explore the relative importance of news media, future research should compare the NAS effects of news media with that of other social actors such as corporations and different interest groups.

As the results illustrate, the proposed concept ‘‘issue ownership network’’ has several practical implications for public relations practitioners and journalists. Drawing upon network theories, the concept suggests an effective way to design political campaign messages by linking different issues to establish the ownership of new issues. When it comes to news coverage, journalists should be more cautious in associating multiple issues when referring to a given candidate. This ‘‘association’’ may impact the public’s perceived ability to handle that issue, even when that competency is not explicitly stated. Overall, the study also shows that a network analysis approach is useful in evaluating campaign or media effects.

The rapidly changing mediascape requires new theories, models, and concepts. This article presents the NAS model and the new concept of issue 574 GUO AND VARGO Downloaded by [162.253.25.113] at 08:44 11 September 2015 ownership network, which allows the analysis of communication effects in a networked information environment. Future studies may consider applying the NAS model to other media and communication concepts such as schema, framing, and priming. Scholars may also use the framework of issue ownership network to predict public opinion in other communication contexts.