ECPR

Install the app

Install this application on your home screen for quick and easy access when you’re on the go.

Just tap Share then “Add to Home Screen”

ECPR

Install the app

Install this application on your home screen for quick and easy access when you’re on the go.

Just tap Share then “Add to Home Screen”

Measuring Political Polarization in Online Social Movement Networks: A Graph Embedding Approach

Political Methodology
Social Movements
Methods
Social Media
Political Activism
Petro Tolochko
University of Vienna
NICOLA RIGHETTI -
Università degli Studi di Urbino
Petro Tolochko
University of Vienna
Annie Waldherr
University of Vienna

Abstract

Political polarization is a significant issue in today’s society, especially in the context of social media, which has become a primary platform for political discourse, opinion sharing, and mobilization. It is crucial to understand the extent and nature of polarization in social networks as it can provide valuable insights into the dynamics of online political communication and its effects on individuals, communities, and the wider society. Several methods have been proposed to measure political polarization in social networks. For example, Barbera (2015) used a distribution of Twitter following links to estimate the ideal points of users, while Yarchi et al. (2021) used network homophily to operationalize network polarization. However, these approaches have limitations, such as requiring a complete network and "anchor accounts" to estimate ideal points or intensive manual content analysis to classify nodes. To address these limitations, this study proposes a novel methodological approach for measuring the ideological positions of users in social networks by using graph embedding techniques. Graph embedding (Grover & Leskovec, 2016) is a method that generates a low-dimensional representation of the social network. This involves using a graph convolutional network to extract features from the graph and generate a low-dimensional vector representation of each node in the network using the node2vec algorithm. By training graph embeddings on a large network, it is possible to capture the underlying relational patterns between nodes. Once the network has been embedded in a low-dimensional space, the nodes’ ideological positions are identified by calculating the cosine similarity between nodes to identify the degree of ideological polarization between these groups. The graph embedding method would allow for the measurement of polarization without the need for a complete graph and would not require intensive manual coding. Using this method, the polarization is operationalized by taking into account the structural properties of a network and is similar to “Interactional Polarization” as defined by Yarchi et al. (2021). As a case study, the paper examines online discourses about environmental movements. Environmental movements have used social media platforms, such as Twitter, to mobilize people around the world and raise awareness about environmental issues. However, these movements have also experienced increasing polarization, with differing opinions and viewpoints leading to conflicts both within the movements themselves and with the wider audience in general. This study analyzes the social media data of environmental movements and examines the patterns of polarization within their social networks. The study aims to answer the following research question: “How polarized are the Austrian and German climate activist networks on Twitter?” The contributions of this paper are twofold. First, it presents a novel methodological approach for measuring polarization in online networks that accounts for the multi-dimensional nature of ideological positions and the structure of social ties. Second, it applies this methodology to investigate the level of polarization in the Austrian/German climate activist networks on Twitter, providing a substantive contribution to our understanding of the nature and extent of polarization in this important context.