Temporal Dynamics of Social Media Impact on Flood Insurance Take-Up From a US County-Level Analysis

This study investigates how social media posts correlate with flood insurance adoption patterns across the United States from 2014 to 2018. Despite increasing flood risks and the widespread availability of flood insurance, insurance adoption rates remains persistently low, prompting the need to understand factors affecting individuals' decisions. By integrating a panel data set combining flood occurrence data, insurance policy information, geo-located social media posts, and community flood reduction measures, we examine how local and national-level social media activity correlates with insurance uptake patterns. Our data-driven analysis includes cumulative linked mixed models with random effects to account for geographical variability across counties. Our findings reveal substantial geographic heterogeneity, captured through county-level random effects, suggesting that local context may filter the influence of social media. Further, while overall social media activity has a modest yet significant effect on insurance uptake, its impact varies depending on timing relative to flood events. Lower-than-average social media activity in a county during disasters correlates with a decreased probability of insurance uptake, while post-flood lower activity shows a positive association with policy adoption, an effect particularly evident following the 2015 Texas-Oklahoma floods. These insights depict a nuanced role of social media in shaping risk perceptions and preparedness, offering valuable implications for risk communication strategies.