论文部分内容阅读
对Poisson分布数据作分析时，如果均数λ不小，可以用正态近似法。与正态分布数据分析相仿，要估计总体均数λ和对假设H0：λ＝λ0或H0：λ1＝λ2作检验时，也需决定样本含量。样本含量n对正态分布数据和Poisson分布数据有不同的含义。在Poisson数据分析中，n是作某种计数的时间或面积，不是观察个体数。对二项分布数据作分析时所需样本含量已在许多统计书中作了介绍，但文献并未对Poisson分布数据分析时样本含量问题作讨论。本文对Poisson分布数据作正态近似分析时所需样本含量的决定，作一些探索和讨论。
When analyzing the Poisson distribution data, if the mean λ is not small, a normal approximation can be used. Similar to the analysis of normal distribution data, when the overall mean λ is estimated and the hypothesis H0: λ = λ0 or H0: λ1 = λ2 is tested, the sample content must also be determined. The sample content n has different meanings for normal distribution data and Poisson distribution data. In the Poisson data analysis, n is the time or area for a certain count, not the number of individuals observed. The sample size required for the analysis of binomial distribution data has been described in many statistical books, but the literature does not discuss the content of samples in Poisson distribution data analysis. This article makes some explorations and discussions on the determination of the sample size needed for Poisson distribution data in a normal approximation analysis.