Entwined Realms: Examining Media Contexts and Information Choice

• In Event: Poster Session: Political Communication 2

Abstract

What prompts 9 million individuals to follow Anderson Cooper on Twitter/X? Some do out of appreciation for his talking points, others do because they absolutely oppose. Still others follow him thanks to platform recommendations based on their location or whether they follow others “like Cooper.” While it is intuitive to believe preferences are the basis of such choices, journalists as popular actors on the platform can also serve to integrate discourses across varied proclivities, prompting questions about how people choose information sources on platforms where they have infinite choice. Thus, understanding the role of contexts and their interplay with personal preferences is crucial, as they can both constrain and unbridle information flows. As social media becomes the dominant source of information for most, our study examines how contextual and preferential factors jointly influence the decision to follow an information source on Twitter.
Hypotheses
We examine 485 Indian journalists and their 12 million unique Twitter followers. Moving beyond an outlet-level approach, we focus on individual journalists instead of news organizations, allowing us to consider even those who are independent. In the aggregate, follower behaviors enable us to discern the mechanisms influencing users' information choices rather than focusing on journalists' production behaviors. Our approach, set within India's democratic and diverse media landscape, sheds light on the dynamics of choice given the media and political environments. We discuss our hypotheses as follows:
1. Mainstream news organizations especially in the nationally dominant languages of English and Hindi pivoted into pro-government reporting in the past decade. Journalists’ parent organizations create recognition shaping audiences’ preference to follow them. We hypothesize that journalists from the same organization are likely to share more followers (H1).
2. Preference for thematic information (hard news vs. entertainment) is well-known to shape choice. We posit that journalists who share greater thematic similarity in their tweets are likely to share more followers (H2).
3. Given the geographic/linguistic diversity in India, we posit that journalists from the same locational context will share more followers (H3).
4. Due to the pro-government shift in mainstream media, digitally native outlets, independent journalists, and fact-checkers have gained popularity, disrupting established information structures. We posit digitally native journalists and traditional journalists sharing followers with each other rather than across media types. (H4)
Method
We transform the bipartite network of journalists connected to their followers into a unipartite projection focused solely on journalists. Here, two journalists are connected if they share at least one common follower. Our strategy includes two stages. First we employ Louvain clustering measure to provide initial insights on the network. Next we employ Additive and Multiplicative Effects (AME) Model to capture nodal, dyadic and triadic dependencies.
Measures
We supplement the journalist-follower relational data with hand-coded information comprising structural and preferential factors which may shape a journalist’s influence. These include the location they report from, their parent organization and whether they are affiliated with digital or traditional media. We use these variables to develop 3 binary matrices where an entry is marked 1 if the journalists belong to the same location, organization, or media type and 0 otherwise. We analyze a year of their tweets to examine message similarity between every pair of journalists. Each matrix becomes a covariate in the model along with the nodal attribute of weighted degree, measuring popularity. We control for journalists’ average likes, retweets and number of tweets per week at the node level.
Results
Initial clustering reveals 5 communities, guided by ideological, linguistic, regional as well as media environment related parameters. Clusters delineate independent anti-establishment journalists, pro-government and non-English language journalists, and elite columnists and editors (from the national capital, New Delhi). Additionally, clusters show journalists based in non-Delhi locations as well as those affiliated with international media. Our main findings follow from the AME model, revealing that individual preference for news themes and organizations are significant but are equally influenced by an information source’s location, popularity and digital/traditional affiliation. While clusters show that ideological segregation exists, the model reveals that it is still shaped by the structures shaping media and political contexts. Our study provides an important theoretical and methodological contribution to the debates surrounding information choice. Thus, we argue that news choice in a social network, even in contexts where institutional roles of state and media are fraught, is still enacted within the constraints of the context.

Authors

• Sakshi Bhalla, uiuc

• Harsh Taneja, uiuc