Aerial view crosswalk Credit Varavin88 Shutterstock

Towards a European Polarisation Observatory


A unique partnership under CIVICA Research aims to examine ideologies and attitudes to issues of public debate.

How do we measure polarisation in Europe’s complex socio-political systems? What is the relation between party politics and online polarisation? What is the role of recommendation algorithms mediating online platforms?

These are some of the driving questions of the research project European Polarisation Observatory (EPO), connecting scholars from CIVICA – The European University of Social Sciences. The project was selected in a call for cutting-edge research proposals under CIVICA Research in early 2021 and began its activities in October 2021.

So far, several works have studied polarisation by analysing political attitudes (for example, stances of individuals on left-to-right scales) held by individuals in groups (Democrat- or Republican-leaning individuals). These studies have centred on questions such as “Is the distance between Democrats and Republicans growing or diminishing?”, or “Are individuals from one group receiving cross-cutting content from the other one, and how does this influence polarisation?”.

These models and questions often considered in the United States, however, do not translate easily to European settings, where political competition is distributed among several parties, and where political stances are organised along several social dimensions, beyond left-to-right scales, in a non-trivial way.

The European Polarisation Observatory aims to create such tools to observe and measure polarisation in digital environments in European countries. The tools will be used to analyse online political competition, media consumption dynamics, and the role of algorithms mediating digital space.

An interdisciplinary European partnership

One of the main goals of the European Polarisation Observatory is to assemble a unique interdisciplinary team of scholars – from sociologists and political scientists to physicists and computer scientists – capable of tackling the particular challenges of analysing complex social and informational digital spaces.

For this purpose, the project counts as partners the Department of Social and Political Sciences at Bocconi University in Italy, the Centre for European Studies and Comparative Politics and médialab at Sciences Po in France, and the Department of Network and Data Sciences at the Central European University in Austria. The partnership also includes the School of Public Policy at The London School of Economics and Political Science in the UK, with the aim of using the project results to inform the future of digital platform policies and regulations.

Addressing complex challenges through novel approaches

Traditionally, people’s opinions on different issues of the public debate have been studied through polls and surveys. But recent advancements in network analysis and machine learning methods have shown that digital behavioural traces (what users share, click, comment, and the social ties they weave online) can be used to mine opinions at a massive scale.

The advancement of these methods provides a model on which to cast the research questions of EPO. By inferring stances and opinions of individuals, and also of media outlets and political figures, the project will position them in geometrical opinion models, in which dimensions act as indicators of positive or negative attitudes towards several different issues of public debate (left-right stances), but also attitudes towards elites, immigration, multiculturalism, and the internationalisation of the economies, among many others.

These methodological advances rely on the conceptualisation of opinions from political psychology, mathematical inference models, and large volumes of data collected and treated by engineering infrastructures at the Sciences Po médialab. To comply with the General Data Protection Regulation (GDPR), project leaders have set up a system that allows for the data to circulate among all partner institutions.

First findings

Despite its recent start, EPO has produced several findings and scientific publications, and its partners have presented in several renowned scientific conferences.

A recent publication in the proceedings of the IEEE conference on Advanced Social Network Analysis and Mining (Ramaciotti Morales et al., 2021) has delivered a proof of concept of the proposed methodology, and has shown the main opinion dimensions that participate in structuring social networks such as Twitter. A case study looked at France where, while left-right political cleavages participate in structuring the Twitter network of friends, opinions towards “globalisation” (European integration, trade openness) have an even stronger influence.

Another article presented at the ACM Conference on Recommender Systems (Ramaciotti Morales & Cointet, 2021) proposed quantitative metrics for network polarisation and showed how different algorithms can increase or decrease it.

In another recently submitted article, several members of médialab produced a landscape of the spread of fake news using this geometrical map of political opinions. Thanks to lists of fact-checked online news articles, researchers were able to assemble a directory of users who have shared misinformation online, and to map their positions in opinion spaces.

Confirming many other recent studies, they found that left-right polarisation plays an important role in predicting if an individual shares fake news. Surprisingly, they also found that other dimensions play an even greater role in explaining who shares misinformation: in France, negative attitudes towards elites have a strong correlation with the probability that an individual has published content labelled as misinformation.

Politically motivated misinformation has been studied so far at large scale in relation with left-right cleavages. The EPO finding is highly relevant because it shows for the first time how other social dimensions of political competition relate to the spread of fake news, connecting massive empirical observations with theoretical hypotheses on what motivates misinformation.

Research on multiple fronts

Currently, the project researchers are deploying different hypotheses and research questions.

Researchers at LSE work on understanding how countries compare in terms of social cleavages online, looking at the political systems of the UK and Poland. Researchers from the Department of Network and Data Sciences at CEU are investigating generative social and algorithmic processes that could give rise to the observed social networks available in the rich data at the disposal of participants. Investigators at the Department of Social and Political Sciences from Bocconi University are examining how the inferred positions of users lend support to different theorised mechanisms driving social interaction. Researchers at médialab and the Centre for European Studies and Comparative Politics at Sciences Po are working on how to determine the “dimensionality” of public political debates, how many dimensions or cleavages are needed to explain - and what are the drivers that might explain - observed polarisation on different issues.

As part of the project, partners plan to organise a conference on European polarisation in early 2023, tackling the particularities and challenges of this socio-political space, and reuniting the researchers from different disciplines who have contributed.


Ramaciotti Morales, P., Cointet, J.P. and Muñoz Zolotoochin, G., 2021. Unfolding the dimensionality structure of social networks in ideological embeddings. In Proceedings of the 2021 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining (pp. 333-338).

Ramaciotti Morales, P. and Cointet, J.P., 2021. Auditing the Effect of Social Network Recommendations on Polarization in Geometrical Ideological Spaces. In Fifteenth ACM Conference on Recommender Systems (pp. 627-632).

Ramaciotti Morales, P. and Muñoz Zolotoochin, G., 2022. Measuring the accuracy of social network ideological embeddings using language models. In 2022 International Conference on Information Technology & Systems (ICITS22).

CIVICA Research brings together researchers from eight leading European universities in the social sciences to contribute knowledge and solutions to the world’s most pressing challenges. The project aims to strengthen the research & innovation pillar of the European University alliance CIVICA. CIVICA Research is co-funded by the EU’s Horizon 2020 research and innovation programme. To stay up to date with CIVICA Research developments and opportunities, subscribe to the newsletter.  

Banner photo: Varavin88 / Shutterstock