The goal WP5 is to develop methods for measuring popular opinion on European integration and obtain the data to apply the methods through systematic, quantitative analysis of social media texts and networks, especially the micro-blogging service Twitter. In particular, our goals are to analyse networks of followership to map the flow of information and to scale actors and their dimensions of difference using network models, and second, to analyse the texts of what is being communicated to determine the content and tone of the public discourse present in social media. After conducting the analysis of Tweets in the context of European Parliament elections and published the findings (“Social media and political communication in the 2014 elections to the European Parliament”), we shifted our focus to the social media about an ongoing monumental event related to this issue, namely the United Kingdom’s withdrawal from the European Union.
For WP 5 (Social media), we downloaded and processed Tweets about Brexit during the campaign period the leading up to the Referendum held in June 2016 and its aftermath. The covered period is from January 2016 to June 2017. The collection of Tweets includes over forty million tweets with more than four million unique users. Through the analysis of the Text of tweets and the follower and re-Tweet networks of Twitter users, we documented numerous results including the ability to estimate user positions on Brexit. We also uncovered networks of relations between the topics discussed on social media, using correlated topic models on the texts of the social media posts.
We disseminated the results in presentations in various venues and a publication. The presentations include a few public events at the institutes such as LSE (January 2017) and Alan Turing Institute in London (November 2017), as long as presentations at academic conferences such as Midwest Political Science Association (April 2017, Chicago USA) and European Political Science Association (June 2017, Milan). A part of our findings is published as a journal article entitled “Predicting the Brexit Vote by Tracking and Classifying Public Opinion Using Twitter Data” (in Statistics, Politics and Policy) focusing on the difference and congruence of two measures of public mood on Brexit measured from Tweets and opinion surveys.
We prepared two manuscripts from the project. One is on the topics discussed in the social media by each side of Leave and Remain throughout the campaign period and their similarities and differences with topics expressed in other modes. Another is about the communication style used by the key Twitter accounts of politicians, campaigner, and media.
Follow our new publications here