Understanding the Effects of Intergroup Interactions on Polarization

What is it about?

Increasing divisions between identity groups threaten social harmony and economic progress. Recent studies highlight a growing trend of polarization, emphasizing the need for a thorough understanding of its temporal patterns and potential mitigation strategies. Researchers have measured polarization by looking at the words people use online. However, word frequency analysis only partially captures the meaning contained in a statement. To address this, we develop a new contextualized-embeddings-based metric that provides a more meaningful measure of how much individuals stick to their group's beliefs and attitudes. We apply this method to estimate polarization in the context of COVID-19-related discourse on Twitter. In particular, we infer the religion of nearly 700,000 Indian Twitter users to understand how different groups interact and how it affects polarization. Our findings indicate that polarization tends to reduce when people from different groups interact more. However, this effect is weaker for those who already have entrenched beliefs. In addition, during communal events, such interactions increase group conformity of minorities and increase polarization. Finally, intergroup interactions amplify polarization among politically active people during political events.

Featured Image

Photo by Charu Chaturvedi on Unsplash

Photo by Charu Chaturvedi on Unsplash

Why is it important?

Our research has important implications. It can help monitor polarization dynamics in real-time. In addition, our measure has broad applicability in various settings. For example, it can help estimate partisan divisions in the government or polarization on other social media platforms where some measure of group identity is known or can be inferred.

Perspectives