论文部分内容阅读
敏感性(灵敏度)试验用于研究不能直接测量的连续变量。一些熟知的统计技术适用于处理定量的(guantal)试验响应。本文中,提出了一个序贯敏感性试验策略和估计方法。该试验方法对于确定响应函数的末端百分率点,比通常使用的敏感性试验方法更有效。效率和精度的提高导致了对中心百分率点(例如50%)进行估计。本方法对许多形式的基本的响应分布是稳健的,它不需要正态分布这个可能的限制性假设和激励步长,并且对初始激励水平的选择不敏感。本试验方法分三部分讨论:(1)试验策略;(2)响应模式;(3)试验数据分析和估计。
Sensitivity tests are used to study continuous variables that can not be directly measured. Some well-known statistical techniques are suitable for handling guantal test responses. In this paper, a sequential sensitivity test strategy and estimation method is proposed. This test method is more effective than the commonly used sensitivity test method in determining the end point percentage of the response function. The increase in efficiency and accuracy has led to an estimate of the percentage point in the center (say 50%). The method is robust to many forms of basic response distribution, it does not require a normal distribution of the possible limiting assumptions and excitation steps, and is insensitive to the choice of initial excitation level. This test method is divided into three parts to discuss: (1) test strategy; (2) response mode; (3) test data analysis and estimation.