Euro CSS

Validating models of opinion polarization

in the digital era

Workshop at the European Symposium on Societal Challenges in Computational Social Sciences

Zurich, September 2, 2019, 9.00-12.30, Room: HG D 5.2

Recent political events such as the election of Donald Trump and Brexit have sparked public and scholarly debate about the effects of online communication on public discourse and the formation of political opinions. Many warn, for instance, that algorithmically created filter bubbles and fake news have contributed to the polarization of political opinions, impeding processes of consensus formation and political decision-making in democratic societies. The literature on social influence dynamics in social networks provides a rich arsenal of analytical and computational models that could be applied to study influence processes in both on- and offline social networks. These models, however, differ in their assumptions about the representation of political opinions, the behavioral preferences of agents, their embeddedness in communication structures, and the ways in which agents exert influence. As a result, existing models often make markedly different predictions about the effects of online communication on collective processes such as opinion polarization, and how such processes would differ from or interact with opinion dynamics in offline contexts.

This workshop will be concerned with approaches to empirically validating the micro-foundations and macro-outcomes of models of social influence. The workshop thus covers experimental studies, approaches from social-network analysis, and the analysis of survey or big data, including data gathered on the Internet. The aim is to provide an overview over critical research questions, methodological opportunities and challenges, and the relative strengths and weaknesses of alternative validation approaches.

The workshop will consist of two parts. In Part 1, a series of short presentations by leading scholars will provide an overview over the key assumptions and predictions of existing models requiring empirical validation. Part 2, will be concerned with alternative empirical approaches, based on presentations covering experimental research, social-network analysis, survey research and approaches drawing from data science. In both parts, there will be ample opportunity to discuss.


09:00 Welcome and opening. Andreas Flache

09:10 Opinion dynamics and social media. Michael Maes

09:40 Validating models of opinion polarization on the micro-level. Marijn Keijzer

10:10 General discussion: state of the art.

10:30 [coffee break]

11:00 Using survey data to validate models of opinion dynamics. Jan Lorenz

11:30 Analyzing polarization and emotion dynamics with social media data. David Garcia

12:00 General discussion: challenges and perspectives.

12:30 [closing]