Opinion polls play an important role in modern democratic processes: they are known to not only affect the outcomes of elections, but also have a significant influence on government policy after elections. Recent years have seen large discrepancies between polls and outcomes at several major elections and referendums, stemming from decreased participation in polls and an increasingly volatile electorate. This calls for new ways to measure public support for political parties. In this paper, we propose a method for measuring the popularity of election candidates on social media using Machine Learning-based Natural Language Processing techniques. The method is based on detecting voting intentions in the data. This is a considerable advance upon earlier work using automatic sentiment analysis. We evaluate the method both intrinsically on a set of hand-led social media posts, and extrinsically – by forecasting daily election polls. In the extrinsic evaluation, we analyze data from the 2016 US presidential election, and find that voting intentions measured from social media provide significant additional predictive value for forecasting daily polls. Thus, we demonstrate that the proposed method can be used to interpolate polls both spatially and temporally, thus providing reliable, continuous and fine-grained information about public opinion on current political issues.
Bibliographical noteFunding Information:
Jane Binner joined the Accounting and Finance Department at Birmingham Business School as Chair of Finance in August 2013. Prior to this she worked as Head of the Accounting and Finance Division at Sheffield Management School and as Reader in Economics at Aston Business School for seven years. Jane has a PhD, MSc, PGCE and BA Hons in Economics from the University of Leeds. She has worked with a number of stakeholder groups such as the Home Office, Experian, the Boots Group plc and Wright Patterson Airforce Base. She brings expertise in analyzing the strategic investment decisions of large enterprises through econometric modelling. Jane has conducted research in econometrics for over twenty years and has extensive academic and commercial experience. Binner has attracted over £1000,000 in external research funding, including awards from the EPSRC/ESRC, the Leverhulme Trust, the National Science Foundation, the Jan Wallander Foundation as well as industrial funding from Boots and Experian. Binner has achieved international recognition for her work on the econometric performance of monetary aggregates and is world leading in her field of financial innovation in the construction of money. Jane has recently been appointed as a visiting professor at the College of Business and Economics Research Centre at the University of Wisconsin, USA and as an INDI Fellow at the Institute for Nonlinear Dynamical Inference. She has four books and over seventy publications in the area of Computational Finance and Economics.
© 2021 Elsevier Inc.
- Behavioural intentions
- Machine learning
- Neural networks
- Political polls
- Social media
ASJC Scopus subject areas
- Sociology and Political Science
- Library and Information Sciences