Combining Survey and Big Data Approaches to Misinformation Research

• In Event: Truth Hurts: New Data, Old Problems in Political Communication Research

Abstract

Existing research suggests that the causes of misperceptions among the public are a combination of individual-level and contextual factors, such as exposure to misinformation, biased processing of accurate information, and the information environment itself. But most of this research is based on laboratory experiments, begging questions about misperceptions in the real world. Drawing on a two-wave panel study capturing the full log of user clicks while browsing the internet, this paper links novel measures of information exposure to survey data on the believability of eight contentious claims spread during the 2017 UK General Election campaign. On average, people grew less likely to endorse incorrect claims by the end of the campaign period. To understand this dynamic, we extracted the text of each news article in our sample, enabling us to identify news stories that featured the claims, from sources that are congruent or incongruent with the content of the claim. Combining this with survey measures of initial belief strength, partisanship, perceived issue salience, and political knowledge, we are able to revisit existing explanations of who and under what conditions are most likely to adjust their belief due to information exposure.

Authors

• Susan A. Banducci, University of Exeter

• Laszlo Horvath

• Daniel Stevens, University of Exeter