基于方差分析的样本有效比较研究

ANOVA-based Research on Comparison of Samples Validity

文章从方差分析视角考察样本有效比较问题,以固定样本、重复样本、二分之一轮换样本以及拼配样本等四类典型代表性的样本为比较对象,通过构造四类样本下的常见估计参数的方差表达式,使用蒙特卡罗模拟方法,从样本观测时期、样本结构、方差比例等角度,研究了四类样本在不同形式估计参数下的最优选择问题。

This paper examines the comparison of samples validity from the perspective of analysis of variance(ANOVA).With the four typical samples as comparison objects,including fixed samples,repeated samples,half of the rotating samples and spliced samples,the paper constructs the variance expressions of usual estimation parameters under the four kinds of samples,and uses the Monte Carlo simulation method to study the optimal selection of the four types of samples under different estimation parameters from the perspectives of sample observation period,sample structure and variance ratio.