“What I tweet is not what I think”: towards a comprehensive multi-version two-way agenda-setting framework

• Main Author: Zhang, Yiyan
• Other Authors: Guo, Lei
• Language: en_US
• Published: 2021
• Subjects: Communication; Agenda-setting theory; Computational social science; Media effects; News consumption; Public opinion; Social media
• Online Access: https://hdl.handle.net/2144/43122

While the traditional agenda-setting theory assumes that a unified issue salience pattern (“the agenda”) will transfer from news media to the public, the emergence of the Internet has challenged this classic communication theory in three ways: by providing two versions of the public agenda (i.e., self-reported issue importance measured by a survey versus social media expressions), by affording two versions of the media agenda (i.e., presented on news websites versus on organizations’ Twitter accounts), and by enabling potential two-way agenda-setting effects. This dissertation aims to construct a multi-version two-way agenda-setting framework via (1) elaborating on the theoretical and practical reasons behind the proposed framework and (2) empirically testing the framework by combining survey and digital texts data around the 2020 US presidential election. The results show an imbalanced two-way agenda-setting relationship, with the traditional media-to-public direction still stronger than the reverse. While the two versions of the media agenda were similar to each other, what people thought was found to be different from what they tweeted. This dissertation also explored the moderating effects of issue-, media-, and individual-level characteristics on the direction and strength of the agenda-setting effects. The issue-wise comparison showed stronger effects in both directions among obtrusive issues, compared to non-obtrusive issues. Interestingly, traditional, non-digital-native media presented a slightly stronger two-way agenda-setting relationship between their news tweets and citizens' tweets compared to digital-native media. This difference, however, was not found in news websites. Individuals with specific characteristics, such as being females, being older, being white, as well as having lower income, lower opinion leadership, and lower social capital, were more likely to influence and be influenced by the media agendas compared to their counterparts. Also, while the well-educated population followed the agenda of news websites more closely, the group with a lower education level followed news tweets on more issues. Finally, the last chapter discusses theoretical, methodological, and practical implications.