Elite Political Discourse has Become More Toxic in Western Countries

TL;DR

The study investigates whether elite political discourse has become more toxic across Western democracies by analyzing nearly 18 million tweets from parliamentarians in 17 countries over five years. It combines Perspective API toxicity scores with zero-shot large language model topic labeling and uses beta regression with country-fixed effects to identify determinants, including party family, government participation, electoral periods, and COVID-19 shocks. The findings show a general rise in toxicity linked to radical-right and opposition status, with 'culture-war' topics (e.g., migration, LGBTQ+ rights) being especially toxic; the COVID-19 period briefly reduced toxicity, and opposition is often more toxic than governing parties. The results imply a troubling normalization of uncivil discourse driven by the media attention economy, suggesting that addressing toxicity requires structural interventions in political incentives and digital ecosystems to preserve constructive democratic dialogue.

Abstract

Toxic and uncivil politics is widely seen as a growing threat to democratic values and governance, yet our understanding of the drivers and evolution of political incivility remains limited. Leveraging a novel dataset of nearly 18 million Twitter messages from parliamentarians in 17 countries over five years, this paper systematically investigates whether politics internationally is becoming more uncivil, and what are the determinants of political incivility. Our analysis reveals a marked increase in toxic discourse among political elites, and that it is associated to radical-right parties and parties in opposition. Toxicity diminished markedly during the early phase of the COVID-19 pandemic and, surprisingly, during election campaigns. Furthermore, our results indicate that posts relating to ``culture war'' topics, such as migration and LGBTQ+ rights, are substantially more toxic than debates focused on welfare or economic issues. These findings underscore a troubling shift in international democracies toward an erosion of constructive democratic dialogue.

Figures (5)

• Figure 1: The evolution of the average level of toxicity by party across all included countries. The figure shows a clear rise in the level of toxicity, nearly doubling from the initial levels. The figure also shows a clear impact of the COVID-19 pandemic, resulting in a brief but substantial fall in the level of toxicity.
• Figure 2: Focusing on the United States, showing the weekly average toxicity by party, highlighting the impact of relevant events such as the impeachment, COVID-19 pandemic, and the 2020 election. The figure reveals shifting inter-party dynamics of toxicity, driven both by external events, and which party is in power.
• Figure 3: (a) The average toxicity of each topic across the included countries. Toxicity values are normalized within each country by computing the z-score of toxicity within each country. After normalization, we calculate the mean normalized toxicity for each topic. (b) The total frequency of topics across all countries in the dataset, normalized by the total number of tweets in the respective country, to ensure that countries with a larger number of tweets do not disproportionately influence the results. (c) Focusing on the case of the United States to examine differences across parties in the level of toxicity associated to specific political issues. The figure shows the mean toxicity of posts by their associated topic for the respective parties. (d) Focusing on the fraction of posts associated to each topic by party in the United States. The figure suggests that less polarized issues tend to be subject of more frequent discussion, and shows substantial differences in the focuses of discussion between the two parties.
• Figure 4: Figures drawn from a single beta regression model, with monthly average toxicity levels by political party as the dependent variable. The model includes party family, party participation in government, election periods, the impact of the COVID-19 shock, a time trend as independent variables. It also includes country fixed effects. (a) reports the predicted level of toxicity by party family, (b) reports the coefficients of the variable government participation, COVID-19 shock and election period, and (c) reports the predicted level of toxicity by month of analyses.
• Figure 5: This is a widefig. This is an example of long caption this is an example of long caption this is an example of long caption this is an example of long caption

Theorems & Definitions (7)

• Theorem 1: Theorem subhead
• Proposition 2
• Example 1
• Remark 1
• Definition 1: Definition sub head
• proof
• proof : Proof of Theorem \ref{['thm1']}