[Paper Review] Opinion dynamics in social networks: From models to data

Antonio F. Peralta, János Kertész+1 more|arXiv (Cornell University)|Jan 4, 2022

Opinion Dynamics and Social Influence33 citations

TL;DR

A survey of discrete and continuous opinion dynamics models, and how they are validated with elections, polls, and controlled experiments to bridge theory and data in social networks.

ABSTRACT

Opinions are an integral part of how we perceive the world and each other. They shape collective action, playing a role in democratic processes, the evolution of norms, and cultural change. For decades, researchers in the social and natural sciences have tried to describe how shifting individual perspectives and social exchange lead to archetypal states of public opinion like consensus and polarization. Here we review some of the many contributions to the field, focusing both on idealized models of opinion dynamics, and attempts at validating them with observational data and controlled sociological experiments. By further closing the gap between models and data, these efforts may help us understand how to face current challenges that require the agreement of large groups of people in complex scenarios, such as economic inequality, climate change, and the ongoing fracture of the sociopolitical landscape.

Motivation & Objective

• Explain the motivation for studying opinion dynamics and how opinions evolve through social interaction.
• Classify and describe key models with discrete and continuous opinions.
• Discuss analytical and computational methods to analyze these models on networks.
• Review empirical data sources and experimental approaches used to validate models.

Proposed method

• Describe transition-rate based discrete opinion models and the role of infection/recovery analogy.
• Present continuous-opinion models including bounded confidence and Deffuant-type updates.
• Discuss mean-field and higher-order approximations, plus connections to other dynamical systems.
• Review data sources including elections, polls, and computerized social network experiments.
• Highlight how model validation informs understanding of consensus, polarization, and fragmentation.

Experimental results

Research questions

• RQ1 What are the main modeling frameworks for discrete and continuous opinions in social networks?
• RQ2 How do different opinion dynamics models compare when tested against empirical election/poll data and controlled experiments?
• RQ3 What mechanisms lead to consensus, polarization, or fragmentation in these models?
• RQ4 How do network structure and external influences shape the outcome of opinion dynamics?
• RQ5 What empirical evidence best constrains or validates the infection/transition-rate forms used in these models?

Key findings

• Discrete opinion models use transition rates to capture imitation, with linear (voter-like) and non-linear (q-voter, threshold) dynamics.
• Continuous opinion models show consensus, polarization, or fragmentation depending on bounded confidence parameter and initial conditions.
• Mean-field and pair-approximation methods can reproduce qualitative behaviors on complex networks and relate to empirical structures.
• Empirical data from elections, polls, and experiments provide patterns that help validate infection-rate shapes and influence mechanisms.
• Controlled experiments reveal diverse influence shapes (inverted S, S-shaped) and the role of leaders or zealots in shaping outcomes.
• Data-driven comparisons reveal that models can capture broad statistical regularities but results depend on context and experimental setup.