摘要
PPS抽样时,一般认为规模变量与目标变量的相关性越强,抽样的精度越高,但少有文献研究相关性的具体内涵以及与抽样精度的内在关联。本文构建了匹配系数反映规模变量与目标变量的比率相关关系,对PPS抽样下的样本匹配系数进行了构建和分析,并通过蒙特卡洛模拟揭示了该统计量的良好性质。本文研究认为,样本匹配系数在抽样精度评价时有着良好的应用效果。
It is generally believed that in PPS sampling, when the auxiliary variable has more correlation with the target variable, the sampling precision will be higher, but there is little research on specific connotation of correlation coefficient and its intrinsic connection with the sampling precision. In this paper, matching coefficient is proposed to reflect the ratio correlation between the auxiliary variable and target variable, then sample matching coefficient based on PPS sampling is built and analyzed, and we show the good properties of the statistics by Monte Carlo simulation. Finally, this paper reveals the sample matching coefficient has good application effect in sampling precision evaluation.
出处
《统计研究》
CSSCI
北大核心
2014年第12期88-91,共4页
Statistical Research
基金
国家社会科学基金项目"城市化进城中农民工的就业歧视及其社会风险研究"(12BJY042)
中国人民大学科学研究基金"适应性整群抽样研究"(14XNH103)的资助
关键词
PPS抽样
规模变量
匹配系数
方差估计
PPS sampling
auxiliary variable
matching coefficient
variance estimation
