Comparing U.S. household disaster preparedness across natural hazard risks: a Hierarchical Bayesian item response theory (IRT) analysis of protective motivation and sociodemographic factors

Understanding how households prepare for different types of natural hazard risks is critical for effective and inclusive disaster risk reduction. This study investigates the variations in U.S. household disaster preparedness across five major risk types based on nationally representative data from the 2023 FEMA National Household Survey (NHS). We develop a “predictability–destructiveness” typology to classify risk types and integrate Protection Motivation Theory (PMT) to examine the cognitive and sociodemographic mechanisms behind. To address the gaps in cross-risk comparison and the limitations of traditional additive measures, we employ a Hierarchical Bayesian Item Response Theory (IRT) model to account for the latent difficulty of specific preparedness actions. The results yield three key insights: (1) Although preparedness actions differ in difficulty, overall preparedness levels remain relatively consistent across risk types. (2) Coping appraisal emerges as a consistent and significant predictor of preparedness (CIs > 0, pds = 1), while the effect of threat appraisal is limited. (3) Sociodemographic factors such as education and income show heterogeneous effects depending on the risk types, particularly for wildfire and riverine flood, while other variables, including gender and rurality, have minimal impact in certain risk contexts. These findings highlight the importance of tailored, risk-specific DRR strategies and suggest the value of integrating interdisciplinary theory with context-specific measurements to assess and support household resilience in multi-hazard environments more accurately.