Evaluation of the performance of stopping rules in sequential analysis

有一些统计推断的问题，没有固定样本容量的程序可用。例如，为置信水平为95%的总体均值构造定宽置信区间的问题与之对应。对于这种情况，我们可以通过随机给定样本大小的顺序抽样方法来解决问题。在本研究中，我们考虑了在未知方差有已知下界的情况下两阶段估计过程的性能评价问题，并建立了一个比以往研究更准确的评价理论。

There are some problems of statistical inference for which no procedures of fixed sample size are available. For example, the problem of constructing fixed-width confidence interval for the mean of a population with confidence level 95% corresponds to it. For this case, we can solve the problem by means of a sequential sampling method where the size of sample is randomly given. In this research, we considered the problem of evaluating the performance of two-stage estimation procedure, provided the unknown variance has the known lower bound, and established a theory which gives more exact evaluation than precedent research.