Freezing-Rain Experience and Household Preparedness: A Bayesian Network Analysis of Disruption Severity and Risk Normalization.

Households may have different preparedness profiles for freezing rain depending on the severity of previous disruptions. This study examines the conditional associations among experience frequency, disruption severity, risk normalization, and household preparedness. Cross-sectional survey data from 533 adult respondents representing individual households in five cities in the Hunan and Hubei Provinces were analyzed using a 12-node Bayesian network integrating disaster experience, risk normalization, risk perception, preparedness knowledge, efficacy beliefs, preparedness constraints, and two preparedness outcomes. Scenario inference, 2000 case-resampling bootstrap replications, repeated stratified 10-fold cross-validation, and parameter sensitivity analysis were conducted. Increasing disruption severity was associated with higher probabilities of risk perception, preparedness knowledge, response efficacy, supplies and outage preparedness, and mobility and care continuity preparedness. Under low disruption severity, repeated experiences corresponded to a 75% probability of high risk normalization, whereas the probabilities of high supplies and outage preparedness and high mobility and care continuity preparedness were 12% and 13%, respectively. Repeated cross-validation showed stronger predictive discrimination for supplies and outage preparedness than for mobility and care continuity preparedness, with mean macro-AUC values of 0.801 and 0.698, respectively; the latter result was interpreted as exploratory. The findings indicate that the preparedness implications of disaster experience depend more on experienced consequences than on exposure frequency alone and support differentiated, calibrated, and context-sensitive preparedness guidance.