COVID-19 Policy Uptake
This project investigtate public sentiments and attitudes toward lockdown policies. In our quest to gauge the risk of human interactions amid the COVID-19 pandemic, we initiated the classification of tweets into pro- and anti-lockdown datasets, focusing on tweets containing pertinent information. Subsequently, we introduced the concept of pro- and anti-lockdown ratios as metrics to gauge the populace’s compliance with pandemic-related regulations. These calculated ratios were then employed in correlation analyses, aligning them with increases in case numbers and the social distancing index, accessible through the COVID-19 Impact Analysis Platform published by the Maryland Transportation Institute. Our findings revealed a negative correlation between the anti-lockdown ratio and the state-level social distancing index, suggesting that regions with higher anti-lockdown sentiment are more likely to witness outbound travel. This approach holds the potential to furnish government bodies and healthcare institutions with insights into public opinions and behaviors during the pandemic, enabling a critical evaluation of policy effectiveness. This study extended the analysis to figure out the associations of state-level anti-lockdown ratio with geodemographic factors. This investigation sheds light on how demographic attributes influence public attitudes to-ward lockdown policies and offers insights into the development of effective and targeted interventions to promote lockdown compliance.
Article
Lingyao Li, Zihui Ma, Hyesso Lee, Sanggyu Lee. “Can social media data be used to evaluate the risk of human interactions during the COVID-19 pandemic?.” International Journal of Disaster Risk Reduction.
The U.S. has taken multiple measures to contain the spread of COVID-19, including the implementation of lockdown orders and social distancing practices. Evaluating social distancing is critical since it reflects the risk of close human interactions. While questionnaire surveys or mobility data-based systems have provided valuable insights, social media data can contribute as an additional instrument to help monitor the risk of human interactions during the pandemic. For this reason, this study introduced a social media-based approach that quantifies the pro/anti-lockdown ratio as an indicator of the risk of human interactions. With the aid of natural language processing and machine learning techniques, this study classified the lockdown-related tweets and quantified the pro/anti-lockdown ratio for each state over time. The anti-lockdown ratio showed a moderate and negative correlation with the state-level social distancing index on a weekly basis, suggesting that people are more likely to travel out of the state where the higher anti-lockdown level is observed. The study further showed that the perception expressed on social media could reflect people’s behaviors. The findings of the study are of significance for government agencies to assess the risk of close human interactions and to evaluate their policy effectiveness in the context of social distancing and lockdown.