Climate Change Adaptation at Multi-levels: Bayesian Belief Networks Approach Links with Policy Implication for the Vietnamese Uplands

Climate change is a global challenge that threatens human well-being and ecosystems through the increasing intensity and frequency of natural hazards. Vietnam, particularly its upland and mountainous regions, is highly vulnerable to climate-related risks; however, climate change awareness remains uneven across local communities. This study examines how upland residents in Yen Bai province perceive natural hazards and how these perceptions shape adaptation intentions. A household survey was conducted in 2017 using a seven-point Likert scale questionnaire structured around an extended Protection Motivation Theory (PMT) framework. Respondents were divided into two analytical groups based on prior awareness of climate change. Bayesian Belief Networks (BBN) were developed separately for each group using structure learning and parameter estimation to identify probabilistic dependencies, key drivers, and control variables influencing preparedness and adaptation intention. The results indicate that ethnicity plays a strong explanatory role only in the low-awareness group, revealing distinct perception and preparedness patterns among Kinh, Mong, Tay, and Thai communities when climate change knowledge is limited. In contrast, ethnicity becomes far less influential among respondents who are aware of climate change, suggesting greater convergence in interpretation and decision-making once awareness is established. The BBN analysis further identifies multi-level leverage points, including individual perceptions and beliefs, household agricultural responses, community preparedness, local development support, land-use practices, and communication pathways, through which policy interventions can enhance adaptation intentions. Overall, the findings demonstrate the value of integrating PMT with BBN to support evidence-based adaptation planning and scenario simulation in Vietnam’s upland contexts.