Using Twitter to Analyze Political Polarization During National Crises

TL;DR

This study analyzes Twitter data to measure how national crises like hurricanes and pandemics temporarily increase political polarization between Democrats and Republicans, revealing a measurable 2.8% spike during such events.

Contribution

It demonstrates that national crises significantly elevate political polarization, providing empirical evidence of increased partisan divide during times of national stress.

Findings

A 2.8% increase in polarization during crises

Polarization spikes are consistent across different types of national events

Twitter sentiment analysis effectively captures short-term political divides

Abstract

Democrats and Republicans have seemed to grow apart in the past three decades. Since the United States as we know it today is undeniably bipartisan, this phenomenon would not appear as a surprise to most. However, there are triggers which can cause spikes in disagreements between Democrats and Republicans at a higher rate than how the two parties have been growing apart gradually over time. This study has analyzed the idea that national events which generally are detrimental to all individuals can be one of those triggers. By testing polarization before and after three events (Hurricane Sandy [2012], N. Korea Missile Test Surge [2019], COVID-19 [2020]) using Twitter data, we show that a measurable spike in polarization occurs between the Democrat and Republican party. In order to measure polarization, sentiments of Twitter users aligned to the Democrat and Republican parties are compared on identical entities (events, people, locations, etc.). Using hundreds of thousands of data samples, a 2.8% increase in polarization was measured during times of crisis compared to times where no crises were occurring. Regardless of the reasoning that the gap between political parties can increase so much during times of suffering and stress, it is definitely alarming to see that among other aspects of life, the partisan gap worsens during detrimental national events.