Sensing the Pulse of the Pandemic: Geovisualizing the Demographic Disparities of Public Sentiment toward COVID-19 through Social Media

Social media offers a unique lens to observe users emotions and subjective feelings toward critical events or topics and has been widely used to investigate public sentiment during crises, e.g., the COVID-19 pandemic. However, social media use varies across demographic groups, with younger people being more inclined to use social media than the older population. This digital divide could lead to biases in data representativeness and analysis results, causing a persistent challenge in research based on social media data. This study aims to tackle this challenge through a case study of estimating the public sentiment about the COVID-19 using social media data. We analyzed the pandemic-related Twitter data in the United States from January 2020 to December 2021. The objectives are: (1) to elucidate the uneven social media usage among various demographic groups and the disparities of their emotions toward COVID-19, (2) to construct an unbiased measurement for public sentiment based on social media data, the Sentiment Adjusted by Demographics (SAD) index, through the post-stratification method, and (3) to evaluate the spatially and temporally evolved public sentiment toward COVID-19 using the SAD index. The results show significant discrepancies among demographic groups in their COVID-19-related emotions. Female and under or equal to 18 years old Twitter users expressed long-term negative sentiment toward COVID-19. The proposed SAD index in this study corrected the underestimation of negative sentiment in 31 states, especially in Vermont. According to the SAD index, Twitter users in Wyoming (Vermont) posted the largest (smallest) percentage of negative tweets toward the pandemic.