Crowdsourced Fact-Checking or Biased Commentary? Analyzing Political Bias in Twitter's Community Notes
Simon Fox Kuuse, Uku Kangur, Roshni Chakraborty, Rajesh R. Sharma · 2025
The rise of social media as a primary source of news has amplified concerns about bias and misinformation, with platforms like Twitter playing a major role in shaping public opinion. Unchecked bias can distort public perception, influence elections, and deepen societal polarization. As traditional fact-checking struggles to keep pace with the flood of online content, crowd-sourced solutions like Twitter's Community Notes offer a promising yet understudied approach to addressing these challenges. This paper provides an exploratory analysis of political bias in Twitter's Community Notes. We analyze 323,382 Community Notes from January 2021 to November 2023, using bias labels based on linked sources. We investigate lexical, sentiment, and temporal features to uncover the bias patterns. Our results show that right-leaning Community Notes have lower readability while left-leaning Notes exhibit more negative sentiment and are more connected to real-world events.