The Dynamics of Affective Polarisation during the 2020 US Presidential Elections

• In Event: Negative Partisanship, Affective Polarization, and Political Divisions

Abstract

Affective polarisation, the deep emotional divide between political groups, poses a significant threat to contemporary American democracy. However, the precise mechanisms by which polarisation spreads and the direction of its influence between politicians and the public remain unclear. While some scholars contend that polarising language originates among the politicians to the public, others argue that it instead emanates from the public sphere and influences politicians. To adjudicate among these competing theories, I analyse a novel dataset of over 9 million tweets posted in the US between 2016 and 2020 by US politicians, news media outlets, and public users. The levels of polarisation in tweets are measured by supervised machine learning algorithms developed by human coders and generative AI. Subsequently, I study the temporal and thematic pattern of polarising language via vector autoregressive models. This paper provides novel insights into the dynamics of polarisation, with potential implications for democratic discourse and civic engagement in the US and beyond.

Author

• Giuliano Formisano, University of Oxford