PHEER Research Brief: Can Social Media Save Lives During Hurricanes? Using Location-Based Social Media Data to Evaluate Disaster Relief Efforts in Houston after Hurricane Beryl

This study examines how social media activity during and after Hurricane Beryl in July 2024 may have influenced the distribution of relief centers and volunteer assistance. With widespread power outages limiting traditional communication channels, platforms like Twitter became critical for sharing real-time updates and requests for help. Given this context, we aim to answer: Do areas with more disaster-related Twitter activity (e.g., posts, reposts, likes, comments) have better access to resources?

We scraped geotagged tweets from July 7 to July 19, 2024 using hashtags such as #HurricaneBeryl, #Beryl, and #Hurricane. Each tweet was then hand-coded to identify referenced locations through text, media, or embedded links. When direct clues were absent from tweets, we examined users' other posts, profiles, or public information from platforms like LinkedIn, Facebook, and Instagram to collect their location during and after the stuck of Hurricane Beryl.

Resource allocation data came from the City of Houston’s Department of Emergency Management, Volunteer Houston, and the American Red Cross. Using Principal Component Analysis (PCA), we created two composite indices to capture the overall resource accessibility and concentration during the Hurricane. We then applied Ordinary Least Squares (OLS) regression, controlling for demographic and socioeconomic factors, to assess the association between social media activity and disaster resource distribution across Houston zip codes.

To our knowledge, this is the first empirical study to empirically examine the association between social media activity and resource allocation from a geographical standpoint. The findings can inform causal analyses and guide researchers, local governments, and communities in improving disaster relief and public health outcomes.