Effects of Base Rate, Environmental Uncertainty, Values, and Feedback on Accuracy and Performance in Selection and Detection Decision Making

许多重要的决定都是基于截止值。例如，医生可能怀疑，但不确定，一个孩子有耳部感染。医生的怀疑是根据孩子的症状做出的判断，如果这种怀疑足够强烈，也就是说，如果超过了某个临界值，医生就会开抗生素。其他“切断”决策包括警察决定逮捕嫌疑人，陪审团决定有罪或无罪，机场安全决定拘留乘客，应急管理人员决定撤离飓风威胁地区，人事经理决定招聘求职者。在每一种情况下，行动或不行动的决定都是基于必须在不确定的情况下做出的判断。决策可能导致两种错误：在不采取行动时采取行动更合适，或者在应该采取行动时不采取行动。这些错误可能代价高昂，而且永远无法完全避免，但了解它们的原因有助于减少它们。这个项目将调查当人们的反馈有限时，他们是如何学会做出决定的。当决策者不知道他们是否做出了正确的决定时，反馈是有限的。通常，决策本身就排除了反馈的可能性。例如，人事经理很少发现没有被录用的申请人是否会成为好员工。这项研究将使用一种新的实验室程序来模拟决策条件。决策情境的重要属性是多种多样的，包括不确定性的数量、利益事件发生的概率、错误和正确决策的成本和收益。这个项目将为研究一类重要的决策建立一个新的范例，将把这个范例应用于人类决策绩效的关键问题，并可能导致改进绩效的方法。

Many important decisions are based on cutoffs. For example, a doctor may suspect, but not be sure, that a child has an ear infection. The doctor's suspicion is a judgment based on the child's symptoms, and antibiotics will be prescribed if that suspicion is strong enough, that is, if it is above some cutoff. Other "cutoff" decisions are police deciding to arrest suspects, juries deciding guilt or innocence, airport security deciding to detain passengers, emergency managers deciding to evacuate a hurricane-threatened area, and personnel managers hiring job applicants. In each case, the decision to act or not is based on a judgment that must be made under uncertainty. Decisions can result in two kinds of error: acting when inaction is more appropriate or not acting when action should be taken. Such errors can be costly and can never be avoided entirely, but understanding their causes can contribute to reducing them.This project will investigate how people learn to make cutoff decisions when their feedback is limited. Feedback is limited when decision makers do not find out if they made the right decision. Often the decision itself eliminates the possibility of feedback. For example, personnel managers rarely find out if applicants not hired would have been good employees. The research will be conducted using a new laboratory procedure designed to simulate decision-making conditions. Important properties of decision making situations will be varied, including the amount of uncertainty, the probability that the event of interest occurs, and the costs and benefits of errors and correct decisions. This project will establish a new paradigm for studying an important class of decisions, will apply that paradigm to critical issues of human decision-making performance, and will potentially lead to methods for improving performance.