Improving the Normal Approximation when Constructing One-Sided Confidence Intervals for Binomial or Poisson Parameters
指出构建二项或泊松参数单侧置信区间时，常用正态近似因分布偏度而误差较大，并给出一种简单有效的偏度校正方法。
We show that the usual normal approximation used during the construction of onesided confidence intervals for binomial or Poisson parameters can be significantly in error due to skewness in the underlying distribution. We provide a simple method of correcting for skewness, and demonstrate that it is effective both in theory and in practice.