Mapping social dynamics on Facebook: The Brexit debate
Introduction
The Arab Spring and Ukrainian revolution showed social media as a liberalizing technology and powerful vehicle of information, engagement, mobilization, able to encourage innovation and democracy. But social media have also changed the way we get informed and form our opinions.
According to a recent report (Newman et al., 2015), approximately 63% of users acquire their news from social media, and these news stories undergo the same popularity dynamics as other forms of online contents (such as selfies and cat photos). As a result of disintermediated access to information and of algorithms used in content promotion, communication has become increasingly personalized, both in the way messages are framed and how they are shared across social networks.
Selective exposure and confirmation bias, indeed, have been shown to play a pivotal role in content consumption and information spreading (Del Vicario et al., 2016). Users tend to select information adhering (and reinforcing) their worldview and to ignore dissenting information (Quattrociocchi et al., 2016, Bessi et al., 2015a, Bessi et al., 2015b, Zollo et al., 2015a). This pattern elicits the formation of polarized groups – i.e., echo chamber – where the interaction within like-minded people might even reinforce polarization (Zollo et al., 2015b, Sunstein, 2002).
Several studies pointed out the effects of social influence online (Centola, 2010, Fowler and Christakis, 2010, Quattrociocchi et al., 2014, Salganik et al., 2006). Results reported in Kramer et al. (2014) indicate that emotions expressed by others on Facebook influence our own emotions, providing experimental evidence of massive-scale contagion via social networks. Recent works (Bessi et al., 2015a, Zollo et al., 2015a), indeed, showed that attempts to debunk false information are largely ineffective. In particular, the discussion degenerates when the two polarized communities interact with one another. The increasing interest in online debates led researchers to investigate many of their aspects, from the characterization of conversation threads (Backstrom et al., 2013) to the detection of bursty topics on microblogging platforms (Diao et al., 2012), to the disclosure of the mechanisms behind information diffusion for different kinds of contents (Romero et al., 2011).
More recently, several doubts about social influence on the Internet have been raised during Brexit – the British referendum to leave the European Union – campaign, where both sides, Leave and Remain, battled it out on social media. Indeed, a big effort has been dedicated to characterize the dynamics of the online Brexit debate, from applying filtering algorithms to study the shape of online data (Stolz et al., 2016), through the investigation of the role of bots on the direction of discussions (Howard and Kollanyi, 2016), to the study of the effects of the referendum result on financial markets (Bianchetti et al., 2016).
In this paper we address the Brexit discussion on Facebook public pages referring to UK based official information sources listed in the European Media Monitor (Steinberger et al., 2013).
Firstly, we characterize the structural properties of the discussion by observing the spontaneous emergence of two well-separated communities; indeed, connections among pages are the direct result of users’ activity, and we do not perform any categorization of contents a priori. Then, we explore the dynamics behind discussion: looking at users polarization toward the two communities and at their attention patterns, we find a sharply bimodal distribution, showing that users are divided into two main distinct groups and confine their attention on specific pages.
Finally, to better characterize inner group dynamics, we introduce a new technique which combines automatic topic extraction and sentiment analysis. We compare how the same topics are presented on posts and the related comments, finding significant differences in both echo chamber and that polarization on the perception of topics. We first measure the distance between how a certain concept is presented on the posts and then the emotional response of users to such controversial topics. Our new metrics could be of great interest to identify the most crucial topics in online debates. Indeed, it is highly likely that the greater the emotional distance between the same concept in two echo chambers, the greater the polarization of users involved in the discussion. Therefore, such a distance may be a key marker to locate controversial topics and to understand the evolution of the core narratives within distinct echo chambers.
Section snippets
Ethics statement
The data collection process was carried out using the Facebook Graph API (Facebook, 2013), which is publicly available. For the analysis (according to the specification settings of the API) we only used publicly available data (thus users with privacy restrictions are not included in the dataset). The pages from which we downloaded data are public Facebook entities and can be accessed by anyone. Users’ content contributing to such pages is also public unless users’ privacy settings specify
Results and discussion
As a preliminary step, we divide all UK based pages in two groups: Brexit pages, that includes those pages engaged in the debate around the Brexit, and Non Brexit pages. Out of 81 pages, 38 posted at least one news story about the Brexit. Hence, we characterize the users behavior on Brexit pages and their related posts.
Conclusions
We address the online discussion around Brexit on Facebook by means of a quantitative analysis on a sample of 5K posts from 38 pages linked to official UK news sources. We observe the spontaneous emergence of two separate communities, where the connections among pages are the direct result of users’ activity and no reference to the shared contents is implied. We further explore the dynamics of the discussion by looking at the polarization of users from the two communities and their attention
References (32)
- et al.
The political blogosphere and the 2004 US election: divided they blog
Powering the New AI Economy
(July 2016)- et al.
Characterizing and curating conversation threads: expansion, focus, volume, re-entry
- et al.
Science vs conspiracy: collective narratives in the age of misinformation
PLOS ONE
(2015) - et al.
Viral misinformation: the role of homophily and polarization
- et al.
Users polarization on Facebook and Youtube
PLOS ONE
(2016) - et al.
Brexit or Bremain? Evidence from Bubble Analysis
(June 2016) - et al.
Fast unfolding of communities in large networks
J. Stat. Mech.: Theory Exp.
(2008) The spread of behavior in an online social network experiment
Science
(2010)- et al.
Finding community structure in very large networks
Phys. Rev. E
(2004)
The spreading of misinformation online
Proc. Natl. Acad. Sci. U. S. A.
Finding bursty topics from microblogs
Using the Graph API, Website
Cooperative behavior cascades in human social networks
Proc. Natl. Acad. Sci. U. S. A.
A comparison of knowledge extraction tools for the semantic web
Bots, # Strongerin, and # Brexit: Computational Propaganda During the UK-EU Referendum
Cited by (182)
Sapling Similarity: A performing and interpretable memory-based tool for recommendation
2023, Knowledge-Based SystemsDe-sounding echo chambers: Simulation-based analysis of polarization dynamics in social networks
2023, Online Social Networks and MediaAmericans misperceive the frequency and format of political debate
2024, Scientific Reports
- 1
These authors contributed equally to this work.