Marginal Comparisons With the Best and the Efficiency Measurement Problem
针对多个总体按未知标量参数排序的问题，构建了单个总体与最优总体参数差异的边际上下界，而非联合界，适用于效率测量等场景。
Consider a set of N populations ranked by an unknown scalar parameter, where a larger parameter value is better. An existing procedure, multiple comparisons with the best, constructs a set of upper and lower bounds for the N – 1 differences between the parameter value of the best population and the other populations, where these bounds hold jointly with a given probability. This article constructs the corresponding upper and lower bounds that hold marginally (for a single given population) rather than jointly, with a given probability.