Applying the Protective Action Decision Model to Household Flood Evacuation: Evidence from the August 2024 Flood in South-central Regions of Bangladesh

Flood evacuation is an effective protective action, yet decisions to evacuate are shaped by complex interactions among risk perception, household characteristics, and contextual constraints. Empirical evidence on household evacuation during rapid-onset riverine floods in Bangladesh remains limited. Guided by the Protective Action Decision Model (PADM), this study examines determinants of household evacuation during the August 2024 flood in south-central Bangladesh. The mixed-methods design integrates a household survey (n = 296) with three focus group discussions and three key informant interviews. Binomial logistic regression identified housing condition, livestock ownership, receipt of early warning information, attachment to property, flood depth, and flood duration as significant predictors of evacuation. Behavioral factors showed the greatest explanatory power (Nagelkerke R² = 0.411; classification accuracy: 75.7–80.4% ), highlighting the role of timely warning dissemination and risk perception in motivating protective action. Qualitative findings revealed persistent institutional challenges — fragmented warning dissemination, inadequate logistics, limited shelter capacity, and weak inter-agency coordination — that constrained timely evacuation despite high public risk awareness. These findings offer evidence for strengthening community-centered evacuation planning, multi-channel early warning systems, and disaster risk reduction strategies in flood-prone, lower-middle-income settings.