One Nation, Divided – A Statistical Analysis of the Impact of Online Rhetoric on Affective Polarization in the United States

Abstract

Politics in the United States has long been characterized by its bipartisan structure wherein voters are often subject to a binary decision between the conservative Republican party and the liberal Democrat party. Though most Americans find themselves in support of a more centrist policy preference, Democrat and Republican elites are diverging ideologically. This divergence has roots dating back several decades and its impact is growing presently. The political landscape of the American government has been the subject of severe scrutiny in the media and become a focal point of online discourse on social media platforms. Consequently, political elites have resorted to Twitter, Instagram, and Facebook to establish their presence online and interact with the public on an unprecedented scale. This study seeks to examine how tweets by United States Senators holding office between 2013 and 2021 impacted polarization levels along this same timeline. Using a psycholinguistic analysis quantifying the underlying tone and subject matter of every tweet from all senators utilizing this platform, a regression analysis was conducted to determine whether or not there is a statistically significant relationship between the language used in tweets and their capacity to incite polarization. Across all explanatory variables and both political parties, it was discovered that tweets with a negative underlying tone, no matter the subject, incite more polarization than positive tweets. In response to these findings, policy recommendations are made to help mitigate the polarizing consequences of tweets from political elites in hopes of fostering a more inclusive environment characterized by meaningful online discourse.