Tracking collective emotions in 16 countries during COVID-19: a novel methodology for identifying major emotional events using Twitter

Abstract

Using messages posted on Twitter, this study develops a new approach to estimating collective emotions (CEs) within countries. It applies time series methodology to develop and demonstrate a novel application of CEs to identify emotional events that are significant at the societal level. The study analyzes over 200 million words from over 10 million Twitter messages posted in 16 countries during the first 120 days of the COVID-19 pandemic. Daily levels of collective anxiety and positive emotions were estimated using Linguistic Inquiry and Word Count's (LIWC) psychologically validated lexicon. The time series estimates of the two collective emotions were analyzed for structural breaks, which mark a significant change in a series due to an external shock. External shocks to collective emotions come from events that are of shared emotional relevance, and this study develops a new approach to identifying them. In the COVID-19 Twitter posts used in the study, analysis of structural breaks showed that in all 16 countries, a reduction in collective anxiety and an increase in positive emotions followed the WHO's declaration of COVID-19 as a global pandemic. Announcements of economic support packages and social restrictions also had similar impacts in some countries. This indicates that the reduction of uncertainty around the evolving COVID-19 situation had a positive emotional impact on people in all the countries in the study. The study contributes to the field of CEs and applied research in collective psychological phenomena.