Nease R F
Department of Internal Medicine, Washington University Medical School, St. Louis, Missouri 63110, USA.
Med Decis Making. 1996 Oct-Dec;16(4):399-403. doi: 10.1177/0272989X9601600410.
Research demonstrates that people violate the independence principle of expected utility theory, raising the question of whether expected utility theory is normative for medical decision making. The author provides three arguments that violations of the independence principle are less problematic than they might first appear. First, the independence principle follows from other more fundamental axioms whose appeal may be more readily apparent than that of the independence principle. Second, the axioms need not be descriptive to be normative, and they need not be attractive to all decision makers for expected utility theory to be useful for some. Finally, by providing a metaphor of decision analysis as a conversation between the actual decision maker and a model decision maker, the author argues that expected utility theory need not be purely normative for decision analysis to be useful. In short, violations of the independence principle do not necessarily represent direct violations of the axioms of expected utility theory; behavioral violations of the axioms of expected utility theory do not necessarily imply that decision analysis is not normative; and full normativeness is not necessary for decision analysis to generate valuable insights.
研究表明，人们违背了预期效用理论的独立性原则，这就引发了一个问题，即预期效用理论对于医疗决策是否具有规范性。作者提出了三个论点，认为违背独立性原则的问题并不像乍看之下那么严重。首先，独立性原则源自其他更基本的公理，这些公理的吸引力可能比独立性原则本身更显而易见。其次，这些公理不必具有描述性才能具有规范性，而且对于预期效用理论对某些人有用来说，它们不必对所有决策者都有吸引力。最后，通过将决策分析比喻为实际决策者与模型决策者之间的对话，作者认为预期效用理论不必完全具有规范性，决策分析也能发挥作用。简而言之，违背独立性原则不一定代表直接违背预期效用理论的公理；预期效用理论公理的行为违背不一定意味着决策分析不具有规范性；而且决策分析要产生有价值的见解并不需要完全具有规范性。
