Social homophily is associated with evacuation destination choice and long-term displacement decisions after the Marshall Fire in the United States

Rapid urbanization and climate change have contributed to a significant rise in the frequency and intensity of disasters, which have resulted in three million adults being displaced from their homes in the United States during 2023, according to the Census Bureau. Using large-scale mobility data, it is now possible to observe and analyze post-disaster mobility dynamics across a longer time period, compared to household surveys, which are often limited in scale. However, much of the mobility data-driven research on evacuation and displacement destination choice behavior has focused on analyzing the effects of physical and spatial factors, often underemphasizing the role of social factors, including individual preferences and social connections. Motivated by the recent literature showing strong associations between social homophily and mobility behavior, we use large-scale data of anonymized GPS traces and online social connections from the Marshall Fires in Colorado, USA, to unravel the associations between social and behavioral factors and evacuation destinations and long-term displacement decisions. We find that behavioral characteristics and social homophily play a significant role in post-disaster mobility decisions. First, incorporating pre-disaster behavior characteristics increases the predictability of evacuated distance by more than fivefold, which is critical for wildfire evacuation. Second, given the same evacuation distance, evacuees chose locations that have a high sociodemographic homophily with their home locations, and locations that have more friendship connections, compared to the level of homophily they are spatially exposed to. The social homophily effect is stronger among White and educated populations than low income, and Black and Asian populations. This effect has significant implications for long-term disaster impacts and policymaking, as it is a significant predictor of displacement and return decisions. Our findings highlight the importance of incorporating the social, behavioral, and economic characteristics to better understand and predict evacuation and displacement dynamics after disasters.