Interplay between Discrete Emotions and Preventive Behavior in Health Crises: Big Data Analysis of COVID-19

Abstract

Understanding the interplay between discrete emotions and COVID-19 prevention behaviors will help healthcare professionals and providers to implement effective risk communication and effective risk decision making. This study analyzes data related to COVID-19 posted by the American public on Twitter and identifies three discrete negative emotions (anger, anxiety, and sadness) of the public from massive text data. Next, econometric analyses (i.e., the Granger causality test and impulse response functions) are performed to evaluate the interplay between discrete emotions and preventive behavior based on emotional time series and Google Shopping Trends time series, representing public preventive behavior. Based on the textual analysis of tweets from the United States, the following conclusions are drawn: Anger is a Granger cause of preventive behavior and has a slightly negative effect on the public's preventive behavior. Anxiety is a Granger cause of preventive behavior and has a positive effect on preventive behavior. Furthermore, preventive behavior is a Granger cause of anxiety and has a negative and lagging effect on anxiety. Exploring how discrete emotions, such as anger and anxiety, affect preventive behaviors will effectively demonstrate how discrete emotions play qualitatively different roles in promoting preventive behaviors. Moreover, understanding the impact of preventive behaviors on discrete emotions is useful for better risk communication.