Health Crises, Community Health Habits, and Pandemic-Driven Food Sentiments: A Twitter-Based Study in the United States

This study analyzes how Twitter users' characteristics (the number of followers and followings) and their context (crisis conditions, community healthy eating, and exercise habits) determine food attitudes during health crises. We analyzed 72,067 US tweets discussing food and COVID-19 from January 20, 2020 to March 20, 2021. The analysis examines tweets mentioning together the topics “food” and “COVID-19” to grasp the sentiments of users regarding food amidst this health crisis. Using sentiment analysis and multilevel modeling, our results show that users with a high number of followings tend to express fewer positive sentiments, whereas the opposite is the case for users with a high number of followers. We found that a worsening of the epidemic conditions increased the propensity to express fewer positive sentiments and that the relationship of community healthy habits with pandemic-driven food sentiments is contingent on these conditions. At low levels of COVID-19 cases, users in healthy-eating communities tend to express positive sentiments, but at high levels, they express more negative sentiments. Regarding community exercise habits, people in these contexts often express fewer positive sentiments. However, as the pandemic worsened, individuals in these communities tended to express more positive sentiments. This study advises policymakers to counter misinformation and develop targeted communication strategies for different communities.