Individual Submission Summary
Share...

Direct link:

Measuring Public Opinion on the Next Generation EU with Twitter Data

Thu, September 15, 4:00 to 5:30pm, TBA

Abstract

This paper develops and tests a method to derive meaningful measures of public opinion from social media data that would be representative of national populations. As a proof of concept, it asks how people in the EU evaluated the Next Generation EU policy from March to November 2020. With Twitter data, it delivers a daily nationally representative measurement of public evaluation of the NGEU for a set of distinct population strata in five EU member states: Germany, the Netherlands, France, Italy, and Spain. With Delvin et al.’s (2018) mBERT language model, the paper identifies a several million-strong pool of policy-related tweets and estimates their stance towards the policy. For every user who expressed their opinion about the NGEU, it estimates their perceived gender, age, region of residence, and education based on users’ Twitter profile information. The demographic information helps estimate the expected propensity to support the NGEU for every distinct population stratum throughout the policy-negotiations period. To adjust for the difference between the Twitter sample and the EU population, strata estimates are reweighted by their known population proportion derived from national census data.

To showcase the usefulness of the measurement, the paper uses the resulting data to explore whether party elite’s cues influence people’s opinion about fiscal integration in the EU. It explores the German government’s – particularly the Christian Democratic Union (CDU) party’s – sudden change in policy preference from opposing any form of a common debt instrument on the European level to endorsing such policy in May 2020. By tracing the difference in opinions on the NGEU among users who follow CDU and CDU politicians on Twitter and those who follow other parties, the paper aims to shed light on the causal relationship between elite cues and public opinion on fiscal integration in the EU.

The paper aims to deliver a twofold contribution: i) it provides panel-level data on distinct demographic strata for scholars to explore relationships between public opinion, elite communication, and EU integration. ii) It adds to political methodology literature by proposing a strategy for deriving representative measures of public opinion from non-representative digital data.

Author