Poster #86 - Multi-Path Cognitive Mechanisms of Typhoon Preparedness: Empirical Evidence Based on Machine Learning

• In Event: Saturday Poster Luncheon

Abstract

Extreme meteorological hazards such as typhoons are increasing in frequency, severity, and spatial extent as climate change intensifies. Active public disaster preparedness has become a key issue in enhancing disaster risk governance capabilities. We develop an analytical framework based on the Protective Action Decision Model (PADM), which integrates theory-driven statistical analysis with data-driven machine learning approach. The results indicate that protective action perception variables are the core driver of public preparedness behavior. Improving information quality and strengthening public efficacy beliefs and social trust are effective in motivating preparedness actions. Machine learning predictions further show that protective action perception variables not only exhibit positive marginal effects, but also that relatively small policy interventions targeting them can trigger system-level disaster preparedness behavioral responses. In addition, the public can be classified into two distinct cognitive profiles. Policy interventions should focus on enhancing self-efficacy and coping efficacy, prioritizing capacity building, operational guidance, and the strengthening of trust mechanisms, supplemented by reducing coping costs, enhancing social trust, and reinforcing responsibility attribution. Overall, our study constructs a multi-level mediation and pathway comparison framework to systematically examine the dominant roles of different cognitive mechanisms across a high-frequency, high-uncertainty, and strong social mobilization typhoon hazard context. Our paper contributions are as follow. First, we focus on a high-frequency, high-uncertainty, and strong social mobilization typhoon hazard context. We not only address the limitation that existing PADM research predominantly focuses on low-frequency or slow-onset risk governance settings, but also extend the applicability of the PADM in risk communication contexts. Second, we systematically examine the differentiated transmission mechanisms through which distinct cognitive pathways influence preparedness behavior by constructing a multi-level mediation and pathway comparison framework and operationalizing antecedent cues as actionable dimensions. We not only answer which cognitive mechanisms assume dominant roles, but also reveal the underlying mechanisms and “black-box” processes linking cues, cognition, and behavior within the PADM framework. Third, we further introduce prediction-oriented machine learning to identify minimum policy intervention intensity, and cognitive structure groupings from both system-level responses. Our study extends the application of the PADM from causal explanation to policy scenario simulation and intervention design.
Keywords:Protective Action Decision Model; Disaster Preparedness Behaviors; Psychological Processes; Machine Learning; Typhoon

Author

• Shuo Mao, Harbin Engineering University
  Presenting Author