Improving the Efficiency of Stochastic Dominance Techniques Using Convex Set Stochastic Dominance
讨论凸集随机占优（CSD）如何减少第二类错误（高效集过大）而不增加第一类错误（排序不准），并用害虫管理策略排序的实例展示CSD将效率集缩小近60%且无需额外偏好约束。
Abstract The advantages of convex set stochastic dominance (CSD) are discussed in terms of extending other stochastic dominance criteria in a way which will decrease Type II errors (large efficient sets) without increasing the Type I errors (inaccurate rankings). An empirical example ranking pest management strategies demonstrates the potential of CSD by reducing the efficiency set by almost 60% without imposing additional constraints on the preference set. It is suggested that CSD may permit more imprecise representations of risk preferences, avoiding utility measurement problems, and still identify efficient sets of acceptable sizes.