Joanna Kopinska
Abstract
We quantify the effects of novax propaganda on vaccines takeup and vaccine- preventable health complications among individuals untargeted by the immunization. We collect the universe of Italian vaccine-related tweets for the period 2013-2018, label the novax stances through Natural Language Processing, and match them with vaccination coverage and vaccine-preventable hospitalizations data at the most gran- ular level available (municipality and year). We then exploit the Twitter network structure as well as the exogeneity of news-related flows of information regarding vaccines to implement a mixed two-stage least squares estimation. We find that a 10 pp increase in the municipality novax stance causes a 0.17 pp decrease in coverage in Measles-Mumps-Rubella vaccine, a 0.8 additional hospitalization every 100k res- idents among individuals untargeted by the immunization (newborns, the immuno- suppressed, pregnant women) and an excess expenditure of 2325 euro, representing a 4% increase in health expenses. Finally, we provide a rationale for the interplay between topic controversialness and polarization in online social networks, showing that if controversy is a byproduct of policy interventions, policymakers should ad- dress these unintended consequences which have the potential to backfire in terms of further polarization and the associated social costs.
Discussant: Sofia Amaral-Garcia (Joint Research Center - European Commission)
Chair: Elisabetta De Cao (Alma Mater Studiorum - Università di Bologna)