The Kavanaugh Hearings: Network Framing Effects in the New Media System

• In Event: Framing Effects in the News

Abstract

The World Wide Web with the attendant growth of social media and the polarization of the electorate has changed the dynamics of information transmission and agenda-setting. In earlier research conducted on the 2004 presidential election we specified an information environment that mirrors and contributes to a polarized political system and developed a methodology that measured those interactions. We studied the flow of opposed claims in three streams of data: (1) web pages, (2) Google searches, and (3) media coverage, in order to determine how false information, which we dubbed factitious informational blends (FIBS), was spread in partisan networks. We found that in its early stages of development Web 2.0 was not sufficient alone for spreading misinformation, but it did influence the agenda of traditional media. We also found no evidence for equality of influence in network nodes. Our findings revealed a tightly knit conservative network of focal influentials who successfully propagated FIBs through message cohesion and counter-FIB rebuttal. The Web, much matured since 2004, now includes a plethora of social media platforms, notably Twitter and Facebook, which mediate discourse and expose publics to an expanded range of polarized networks. This changed environment leads to research questions on the identity of information sources that propagate polarizing issue frames and the prominence of their role in intermedia agenda setting.

In the proposed research--a case study of the confirmation hearings of Supreme Court nominee Brett Kavanaugh and his accuser, Christine Ford—we turn our attention to network framing effects. We specifically advance research questions of 1) Frame identification, 2) Frame amplification, and 3) Frame transfer in order to determine how this controversial issue was covered in partisan networks. To extend our prior approaches, our current methodology responds to the maturation of the Web, changes in search engine optimization, and the rise of Twitter as an influential node in the present information system. As in the prior study we seek to understand how partisan publics are exposed to issue frames on the Kavanaugh controversy in web pages turned up in the first three pages of Google searches, mass media coverage, and tweets.

Our methodology responds to changes in the way Google optimizes its search results by creating three experimental search histories (conservative, centrist, and liberal/progressive) based on partisan web-site preferences. The logic of the algorithm behind Google search, the incentives for savvy web authors, and the reluctance of citizens to dig deeply into search results make the web an efficient vector for the spread of politically motivated information. It is an open question, however, whether the web’s potential for increased participation and influence is equally distributed.

In seeking to build these search histories, we establish these three patterns based on partisan preferences by clearing Google search history in three computers, and then by repeatedly navigating to sites favored by conservatives (e.g., Foxnews.com, Brietbart, the Daily Caller, Washington Times), centrists (e.g., BBC, CBS, USA Today), and liberals (e.g.,Daily Kos, Huffington Post, Raw Story and Blue Nation Watch). For social media streams, we collect tweets for the month-long period that Google Trends reveals as the lifespan of the controversy: September 15 through October 20, 2018.

Our research seeks to determine whether right-leaning networks maintain frame purity more effectively than left-leaning networks as our earlier findings suggested and whether the Web influences framing on mainstream media. Here we include the most widely circulated newspapers (USA Today, NYT, WSJ, LA Times, Chicago Tribune) and the evening news on ABC, current ratings leader. We also seek to determine the role of social media in seeding, amplifying and further propagating these frames. The latter will be examined by comparing the frames emerging in tweets with that of the Web and mass media.

Our study incorporates a range of methodological big data innovations beyond manual frame analysis to include automated frame detection via Python scripts and semantic network analysis. These methodological advances will enable frame detection across all three sources (Web, mass media and social media) to benefit from a synthesis of computerized and manual methods. To establish intermedia agenda-setting with emergent frames, we use time series statistical analysis, specifically Granger causality, to determine whether the time series of any given source X predicts another source Y over and above Y’s ability to predict itself. The choice of Granger causality time series analysis will enable us to capture the dynamic process of frame setting through time lags of one hour or less—a speed which best reflects the rapidity of information diffusion in the altered social media ecology.

Authors

• Andrew Rojecki, University of Illinois, Chicago

• Sharon Meraz, University of Illinois at Chicago