Micro‐Behavioral Responses for Enhancing Urban Flood Resilience: Stated Preference Analysis of Residents' Evacuation Choices in High‐Density Urban Areas

ABSTRACT The increasing frequency of extreme flood events in recent years has posed substantial challenges to the emergency evacuation capacity of urban areas. Existing evacuation plans are largely based on rational assumptions, overlooking the complex decision‐making mechanisms through which residents simultaneously weigh multiple attributes in real‐world scenarios. This study focuses on residents of Beijing and employs a stated preference experiment together with a conditional logit model to examine their preferences regarding evacuation distance, bottleneck risk, traffic control, and organizational mode. It further explores the heterogeneous effects of gender, age, education level, and length of residence. The results reveal that, in the face of flooding, residents' evacuation decisions are driven by survival probability rather than by minimizing travel time. The combination of intelligent signals and government‐planned routes is the only attribute consistently preferred across all groups. However, significant divergences exist among different population segments in terms of bottleneck aversion, organizational mode preference, and sensitivity to distance. Scenario analyses based on the model results indicate that the deployment of intelligent signals and government‐provided guidance yield the highest universal benefits. Dynamic management of high‐risk bottlenecks and community‐based adaptations tailored to elderly groups can effectively enhance evacuation performance. These findings challenge the conventional view of homogeneous evacuation decision‐making among residents and highlight marked differences in risk avoidance choices across population groups, providing empirical support for shifting emergency management from persuasion‐based strategies to differentiated resilience approaches.