Beyond Polarization: Opinion Mixing and Social Influence in Deliberation

2026-01-20

AI Summary

Traditional polarization metrics struggle to capture the dynamic process of opinion reconfiguration during deliberation, particularly the evolving convergence of individual viewpoints. This study introduces a novel dimension—“opinion blending”—and employs Kendall’s rank correlation to measure shifts in opinion rankings before and after deliberation. Drawing on a large-scale online deliberation experiment across 32 countries, comprising 6,232 discussion statements coded with LLM assistance and multi-wave survey data, the research elucidates the mechanisms through which social influence and argument quality shape collective opinion change. Findings reveal that deliberation significantly reduces consensus in opinion rankings across 97% and 93% of issues examined. Moreover, high-quality—not merely novel—arguments robustly predict opinion shifts, thereby overcoming the limitations inherent in conventional variance-based polarization measures.

Technology Category

• Natural Language Processing: Discourse, Pragmatics & Argument Mining
• Knowledge Representation and Reasoning: Argumentation
• Multiagent Systems: Agreement, Argumentation & Negotiation

Application Category

• Social Networks and Social Media: Media and governance, opinion dynamics, filter bubbles, polarization
• Web Mining and Content Analysis: Sentiment analysis and opinion mining
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

Deliberative processes are often discussed as increasing or decreasing polarization. This approach misses a different, and arguably more diagnostic, dimension of opinion change: whether deliberation reshuffles who agrees with whom, or simply moves everyone in parallel while preserving the pre-deliberation rank ordering. We introduce \opinion mixing, measured by Kendall's rank correlation (\tau) between pre- and post-deliberation responses, as a complement to variance-based polarization metrics. Across two large online deliberative polls spanning 32 countries (MCF-2022: n=6,342; MCF-2023: n=1,529), deliberation increases opinion mixing relative to survey-only controls: treatment groups exhibit lower rank correlation on (97%) and (93%) of opinion questions, respectively. Polarization measures based on variance tell a more heterogeneous story: controls consistently converge, while treated groups sometimes converge and sometimes diverge depending on the issue. To probe mechanisms, we link transcripts and surveys in a third event (SOF: (n=617), 116 groups) and use LLM-assisted coding of 6,232 discussion statements. Expressed support in discussion statements strongly predicts subsequent group-level opinion shifts; this correlation is amplified by justification quality in the statements but not by argument novelty. To our knowledge, we are the first to observe how different notions of argument quality have different associations with the outcome of deliberation. This suggests that opinion change after deliberation is related to selective uptake of well-reasoned arguments, producing complex patterns of opinion reorganization that standard polarization metrics may miss.

Problem

• opinion mixing
• polarization
• deliberation
• social influence
• rank correlation

Innovation

• opinion mixing
• Kendall's tau
• deliberative polling
• argument quality
• LLM-assisted coding