Abstract
Vaccines and climate change have much in common. In both cases, a scientific consensus contrasts with a divided public opinion. They also exemplify coupled human–environment systems involving common pool resources. Here we used machine learning algorithms to analyze the sentiment of 87 million tweets on climate change and vaccines in order to characterize Twitter user sentiment and the structure of user and community networks. We found that the vaccine conversation was characterized by much less interaction between individuals with differing sentiment toward vaccines. Community-level interactions followed this pattern, showing less interaction between communities of opposite sentiment toward vaccines. Additionally, vaccine community networks were more fragmented and exhibited numerous isolated communities of neutral sentiment. Finally, pro-vaccine individuals overwhelmingly believed in anthropogenic climate change, but the converse was not true. We propose mechanisms that might explain these results, pertaining to how the spatial scale of an environment system can structure human populations.
Highlights
IntroductionDespite a strong scientific consensus that vaccines do not cause autism and that anthropogenic climate change is real, public debate on both topics continues to polarize many populations (Cook et al, 2013; Taylor et al, 2002)
Despite a strong scientific consensus that vaccines do not cause autism and that anthropogenic climate change is real, public debate on both topics continues to polarize many populations (Cook et al, 2013; Taylor et al, 2002). These public debates are well represented in online social media (Betsch et al, 2012; Love et al, 2013; Newman, 2017; Nicholson and Leask, 2012; Schafer, 2012), including Twitter (Love et al, 2013; Newman, 2017)
Datasets on vaccines and climate change were collected through the Twitter Application Programming Interface (API) using a combination of the passive listener and active searches, satisfying the same search terms as the GNIP datasets (Table 1, the “collected” datasets)
Summary
Despite a strong scientific consensus that vaccines do not cause autism and that anthropogenic climate change is real, public debate on both topics continues to polarize many populations (Cook et al, 2013; Taylor et al, 2002) These public debates are well represented in online social media (Betsch et al, 2012; Love et al, 2013; Newman, 2017; Nicholson and Leask, 2012; Schafer, 2012), including Twitter (Love et al, 2013; Newman, 2017). Both the total volume of market-related tweets and the sentiment of those tweets are correlated to stock market movements (Ranco et al, 2015) These studies suggest that analyzing public Twitter debates on climate change and vaccines could be helpful for understanding real-world decision-making concerning vaccines and climate change, such efforts are still in their early stages (Charles-Smith et al, 2015). The impact of real-world developments in natural systems such as global temperature anomalies are reflected in online social media data (Moore et al, 2019)
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
