From tweets to tensions: exploring the roots of political polarization in Turkish constitutional referendum

Abstract

ABSTRACT This study examines between-group and within-group polarization in online social networks to uncover latent factors and ideological changes over time. The study used 2017 Turkish constitutional referendum as a case focusing on Twitter dynamics between the yes-vote and no-vote communities, examining divergence of worldviews and partisan responses. We examined how changes in partisan asymmetries over time shape echo-chambers in opinion, network, and user polarization providing evidence of polarization at group, sub-group, and individual levels. Findings show differences in polarization between groups at different time periods. Before the referendum, yes-voting users are main drivers of polarization and more inward-looking than no-voting users. In contrast, considering the gap between winners and losers of the referendum and associated mood swings, in the days following the referendum no-voting users are main drivers of polarization. This demonstrates fine-grained details of polarization and changes in users’ sentiment as a latent factor behind seasonal behavior differences observed between two groups. Overall, yes-voting community was ideologically more coherent with a relatively denser network structure and more polarized body of users.