Our work is only possible through the generosity of private philanthropy. Find out how you can support our mission and can contribute to our work.
Join the HJS mailing list and keep up to date.
By Jacob Kishere
On Tuesday 27th of June the Henry Jackson Society and Kevin Hollinrake (MP) welcomed Professor Kenneth Benoit, Professor of Quantitative Social Research Methods, and Head of the Department of Methodology at the London School of Economics and Political Science. Professor Benoit came to present the findings on his recent pioneering research on Brexit and Social Media.
With and extensive grounding in qualitative analysis and in collaboration with colleagues at Imperial College London and elsewhere in Europe, Benoit’s project collected an extensive amount of data analysing as many as 35 million tweets in a 6 month period. His conclusions and data analysis presented were based 26 million tweets collected through hashtags, usernames and common search terms.
Professor Benoit stressed throughout his presentation the sheer vastness of the data pool that twitter provides today with between 400 and 450 million tweets published per day as well as 20% and 29.2% of British and American respective public’s holding an account.
Through a process of supervised machine learning the LSE’s team’s smaller data set was able to map onto the wider data pool with a predictive accuracy of 80-90%.
Professor Benoit presented a number of predictions which broadly re-enforced people’s concerns about significantly differing priorities, strategies and concerns of the Leave and Remain camps. Twitter accounts associated with Leave consistently produced a higher quantity of tweets throughout the majority of the campaign. By a process called ‘Sentiment Analysis’ the LSE team was able to classify tweets by their association with rewarding, positive, negative and quantitative language as well as by themes of politics, power and past or future orientation. They found the Leave campaign to have a far higher quantity of tweets pertaining to ‘rewards’ from leaving the EU. The Leave tweets were characterised by a greater degree of positive language, reward language and emphasis on the future. Conversely, the Remain tweets language was consistently more tentative and more quantitative with emphasis on figures.
Additionally, Professor Benoit used ‘Topic model cluster methods’ to determine the distribution and focus of topics associated with each side of the referendum campaign. Topics featuring highly amongst leave tweeters included: Trump, Islam, taking back control, Jo Cox’s murder, xenophobia, Immigration, Identity and Sovereignty. In contrast, Remainer’s emphasis was far more on the youth, the economy, Scotland and Northern Ireland, the NHS and distinctly; anger against the conduct of the leave campaign. Broadly, Remain tweeters were characterised more by provision of information where leave campaigners were more so by Narrative.
During the Q&A attendees raised concerns over the vulnerability of new media to foreign influence and political extremism. Professor Benoit remarked on the ‘great irony of the information age’ that whilst the information age effectively undermined the tight control of information on which fascism is contingent on, it is entirely vulnerable to populism. That the algorithms for Facebook and twitter, like those for sites like Netflix, effectively exaggerate user’s tendencies-something problematic in the context of political echo chambers. With such a vast data pool Professor Benoit emphasised that there remained much more research to be done.