Parts | Days | Selection | Search | Updates | Downloads | Help

SOE: Fachverband Physik sozio-ökonomischer Systeme

SOE 11: Polarization

SOE 11.2: Talk

Wednesday, March 11, 2026, 16:00–16:15, GÖR/0226

Dynamics of ideologically polarized social media users — •Tristram Alexander and Jamie Tyler — University of Sydney, Australia

Social media users have been shown to exhibit different types of behaviour and preference on social media, depending on their ideological position on a left-right axis. This work investigates whether observed differences may be attributed to intrinsic differences in the users, or whether the differences emerge based on interactions in the social media environment.

The study is based on interactions between users on the Twitter/X platform. Users are classified on a left-right scale based on their retweet activity, and placed into categories based on their position on this scale. The reply activity between users is then examined to determine the rates of interaction between the different classes of user. The population of the users in each interaction class is then estimated, based on frequencies of interaction between the classes.

The study finds that users appear to have the same intrinsic behaviour, irrespective of ideological class, but that they behave differently due to the population imbalance between the classes on the platform. This leads right-leaning users to be more active then left-leaning users, which is posited as a reason for greater observed departure rates of right-leaning users from the platform. A model of this interaction behaviour is developed and mapped to the observations. The results of this work may inform the further development of opinion dynamics models.

Keywords: Polarization; Twitter/X; Conversation dynamics; Opinion dynamics; Hidden parameter estimation