一种更强的经典置信限定义
高能物理 - 实验
2007-05-23 v3
摘要
本文提出了一种在经典统计学中定义置信限的新方法。这是对 Neyman 原始方法的自然扩展，并且具有一个理想的性质：在统计推断中仅使用与问题相关的信息。其结果是对允许的置信带施加了严格的限制，在完全普遍性上排除了诸如空置信区域或不稳定解之类的病态情况。该方法完全通用，可直接应用于所有极限问题。
引用
@article{arxiv.hep-ex/9912048,
title = {A stronger classical definition of Confidence Limits},
author = {Giovanni Punzi},
journal= {arXiv preprint arXiv:hep-ex/9912048},
year = {2007}
}
备注
21 pages, RevTeX. Fixed a trivial mistake in formula (8). This has no consequence on anything else, since its purpose was just to show a visual similarity to formula (1). The bug was never present in the file posted on the INFN preprint server