Abstract
With the rising use of social media in political communication, elected officials often have to respond instantaneously to the demands of their constituents in digital campaigns, even at the expense of compromising partisanship. However, is such eclecticism still valid at the legislative level? This study argues a deceptive signaling effect of Twitter political campaigning in response to the Black Lives Matter (BLM) movement. When the BLM ignited a nationwide demand for police reform in the United States in May 2020, the legislators, though most tweeted to chastise the murder of George Floyd, disappointed the public by killing two bills addressing police brutality. We tested the argument by collecting and examining 541 thousand tweets of the 116th U.S. Congress since 2020 and their voting records. We devised a novel deep-learning classifier model to identify the political attitude of tweets regarding racial justice and policing. We discovered an active bipartisan discussion of racial justice on Twitter momentarily. However, in a month, the responsiveness of the Republicans plummeted, while the Democrats kept the momentum. Moreover, after controlling the effects of BLM protests or politician partisanship, tweeting responses from both parties become insignificant to their voting decision. We conclude that when facing social movement, a bipartisan strategy on social media is adopted by legislators to signal their responsiveness to social demand, but the strategy is transient and ineffective in changing legislative decisions. Namely, Congress members merely “cheap talk” on social media to signal their willingness to engage their voters.
Presenters
Zhaozhi LiStudent, Student in Political Science and Data Science, Duke Kunshan University, Jiangsu, China Charles Chang
Duke Kunshan University Chenglin Zhang
Undergraduate, Data Science Major, Natural Science Dept., Duke Kunshan University, Jiangsu, China
Details
Presentation Type
Paper Presentation in a Themed Session
Theme
KEYWORDS
Signaling Effect, Twitter, BLM, U.S. Congress, NLP, Deep Learning
