Sanderson H, Stahl C H, Irwin R, Rogers M D
University of Guelph, Centre for Toxicology, Ontario, Canada.
Water Sci Technol. 2005;52(6):73-9.
Quantitative uncertainty assessments and the distribution of risk are under scrutiny and significant criticism has been made of null hypothesis testing when careful consideration of Type I (false positive) and II (false negative) error rates have not been taken into account. An alternative method, equivalence testing, is discussed yielding more transparency and potentially more precaution in the quantifiable uncertainty assessments. With thousands of chemicals needing regulation in the near future and low public trust in the regulatory process, decision models are required with transparency and learning processes to manage this task. Adaptive, iterative, and learning decision making tools and processes can help decision makers evaluate the significance of Type I or Type II errors on decision alternatives and can reduce the risk of committing Type III errors (accurate answers to the wrong questions). Simplistic cost-benefit based decision-making tools do not incorporate the complex interconnectedness characterizing environmental risks, nor do they enhance learning, participation, or include social values and ambiguity. Hence, better decision-making tools are required, and MIRA is an attempt to include some of the critical aspects.
定量不确定性评估和风险分布正受到审视,当未充分考虑I型(假阳性)和II型(假阴性)错误率时,对零假设检验提出了重大批评。本文讨论了一种替代方法——等效性检验,它在可量化的不确定性评估中具有更高的透明度和潜在的更多预防措施。鉴于在不久的将来需要对数千种化学品进行监管,且公众对监管过程的信任度较低,因此需要具有透明度和学习过程的决策模型来管理这项任务。自适应、迭代和学习型决策工具及过程可以帮助决策者评估I型或II型错误对决策选项的重要性,并可以降低犯III型错误(对错误问题给出正确答案)的风险。基于简单成本效益的决策工具既没有纳入表征环境风险的复杂相互联系,也没有促进学习、参与,也没有纳入社会价值观和模糊性。因此,需要更好的决策工具,而MIRA就是试图纳入一些关键方面的一种尝试。