以結構方程模式探討台灣地區堰塞湖災害預警與居民認知影響避難決策之研究

• Main Authors: 林宏立, Lin, Hung Li
• Language: 中文
• Published: 國立政治大學
• Subjects: 堰塞湖; 災害預警; 風險溝通; 決策行為; 結構方程模式; Landslide Dam; warning mechanism; risk communication; decision making; Structural Equation Modeling(SEM)
• Online Access: http://thesis.lib.nccu.edu.tw/cgi-bin/cdrfb3/gsweb.cgi?o=dstdcdr&i=sid=%22G0098257008%22.

為有效減少因不可預期的極端氣候災害所產生的傷亡，可透過提升災害預警機制與災害認知使民眾具備更佳的災害應變決策能力。本研究選擇以台灣地區仍不常見的堰塞湖災害為研究對象以呼應極端氣候不可預期的特性，首先整理出居民在進行災害應變決策時的思考模式與程序，並釐清影響最後決策的各項因素，作為本研究主要研究架構的初擬參考；在分析工具方面則選用結構方程模式(Structural Equation Modeling, SEM)，以同時解決將心理層面認知量化與各變項間直線迴歸關係的問題。 為將以上構想付諸實行，本研究先經由文獻回顧提出居民的災害應變決策概念架構，以此概念架構為基礎進行問卷設計，並在台東縣嘉蘭村與高雄市瑪雅里進行問卷調查作業，取得資料後再透過結構方程模式建立適合台灣地區民眾的堰塞湖災害決策模型；最後則針對本研究所建立的模型與分析，對現況風險溝通與預警機制提出相關政策建議。 在試圖達成前述目的的研究過程中，本研究發現台灣民眾在面對不熟悉的堰塞湖災害時，外在的預警訊息、過去的受災經驗，以及家戶狀況等三項因素對決策影響最大；另外雖然本研究所提出的決策架構在台東與高雄兩處受測範圍內均可適用，但仍會因聚落的受災經驗、居民屬性，以及交通區位等特性上的不同而在變項重要性上有所差異。最後，本研究提出相關風險溝通的策略建議，作為未來在面臨堰塞湖災害或其他極端氣候時的政策參考。 === In order to reduce the unpredictable and extreme weather disasters’ casualties effectively, we can improve disaster warning mechanisms and disaster awareness so that people have decision-making capacity for better disaster response. This study selects the Landslide dam’s disaster, which is not common in Taiwan, as the research object. First sorted out the residents’ disaster response policy and procedures of thinking, and clarified the factors affect the final decision as the beginning research framework of this study. Analysis tools are used in structural equation model (SEM), to address both the psychological aspects of cognitive variables to quantify and the question of the relationship between the linear regressions. On the research process, first put forward the residents’ conceptual framework for disaster response decision-making through the literature review, and using this conceptual framework as the basis for questionnaire design. Taitung County and Kaohsiung City are the questionnaire surveys in this study. After the data obtained, we established a suitable decision model for Taiwan’s Landslide dam disaster through the SEM tool. Finally, we can make recommendations on risk communication policy and early warning mechanism through this model. In the research results, this study found that the external warning message, past disaster experience, and household conditions are three of the most important decision factors when Taiwan public is facing the Landslide dam disaster. In addition, although this framework can be applied in Taitung and Kaohsiung areas, but there will still be differences because of the differences in some characteristics such as the affected experience, residents’ properties, and location of traffic. Last, this study proposes some strategies of risk communication as the policy reference when facing the Landslide dam’s disaster or other extreme weather disasters in the future.