Abstract

Prioritization is widely used in quality practice. Since the process is accompanied by unavoidable errors, in order to improve accuracy, prioritization is usually repeated by M independent experts, followed by aggregation of their decisions. The article studies the bias of such a procedure, under the assumption of a “gold standard” existence and Mallows’ model of errors creation, for different combinations of M, number of alternatives—n and probability of error-free prioritization P₀. Thereby it becomes possible to know how many experts are required and how accurate must they be in order to prevent the bias from exceeding some threshold.

中文翻译：

---

Mallows错误产生模型下多专家优先排序的准确性

摘要

优先级在质量实践中被广泛使用。由于该过程伴随着不可避免的错误，为了提高准确性，通常由M 个独立专家重复优先级排序，然后汇总他们的决策。文章研究了这种程序的偏差，在“金标准”存在和 Mallows 错误产生模型的假设下，对于M 的不同组合、备选方案的数量 - n和无错误优先级的概率P ₀。因此，可以知道需要多少专家以及他们必须有多准确才能防止偏差超过某个阈值。