建構RCA 事件調查的因果推理──結構因果與模型逆推法的取向

Constructing a Causal Reasoning for the RCA Event─An Approach based on Structural Causation and Abduction

本文應用科學方法論最新發展的「模型逆推法」來回答一個科技與社會的爭議：曾在美國無線電公司任職的女性勞工罹患癌症，是否是因為該廠使用三氯乙烯、四氯乙烯等有機溶劑所造成的？台灣的毒理學家、公衛學家和流行病學家已針對上述和其他相關的問題發表九篇科學論文，但其多數結論模棱兩可，互不相容——因為流行病學傳統使用的統計調查方法所預設的因果觀念是「律則性的因果觀念」，這個因果觀念和它相關方法學很難回答「獨一無二」、「資料不全」事件發生的原因，也不能完整考慮此問題發生的社會脈絡。筆者嘗試發展出一個「結構因果觀念」，並結合「模型逆推法」，應用到這個問題上，由此推出一個假設。本文比較這個推論得到的結論，並與使用流行病學調查方法的結論做比較；再使用現有流行病學調查得到的統計資料，以檢驗「模型逆推法」推出的假設。全文的結論是，「曝露於有機溶劑是RCA女性勞工罹癌的原因」是一個比相反結論更好、更可靠的假設。

This paper develop a new pattern of reasoning to solve a science, technology and society (STS) problem: Were the organic solvents such as trichloroethene and tetrachloroethene the factors that caused the cancer incidence of female workers who had worked in Radio Company of America (RCA) during the 1970s? Up to 2009, nine toxicological and epidemiological papers addressing to this and other derived problems had been published in different international journals. Most of those papers suggested uncertain, ambiguous, and incompatible answers. In this paper, I analyzes that the statistical methodology presupposed by those epidemiological papers implies the regularity notion of causation. I argue that this notion cannot solve the RCA problem. To get a reliable and convincing solution, I propose a notion of structural causation with the abduction based on models. According to the reasoning pattern, “the organic solvents were the causes of the cancer incidence of female workers in RCA” is a better hypothesis than its negation.