不等群下的整群抽样问题研究  

Cluster Sampling Out of Clusters of Unequal Sizes

作　　者：陈仁恩[1] 卢宗辉 

机构地区：[1]厦门大学计划统计系,福建厦门361005 [2]深圳国信证券综合研究所,广东深圳518060

出　　处：《厦门大学学报（哲学社会科学版）》2006年第5期107-112,共6页Journal of Xiamen University(A Bimonthly for Studies in Arts & Social Sciences)

摘　　要：在社会经济领域中,群大小不等是常见的现象。在群大小不等条件下进行整群抽样最主要的问题是样本群大小的随机变动。样本群大小的随机变动会给抽样调查带来种种不利的影响。要提高整群抽样估计的准确度,就必须控制样本群大小的变动。在群大小不等条件下,简单随机抽样无偏估计方法是不可取的,应尽可能地采用麦多方法、PPS方法或简单随机抽样比率估计方法,以提高整群抽样效率。另外,改进总体推估方式也是提高整群抽样效率的简便方法,应值得高度重视。In the soeio-economic field, clusters of unequal sizes （COUS） are a common phenomenon. The main problem with cluster sampling out of COUS is the random change of sizes, which may affect the accuracy of sampling. It follows that we must control the change of sizes of sample clusters in order to enhance the accuracy of cluster sampling estimation. It is argued that simple random sampling unbiased estimation method is not advisable in the case of COUS, that the Madow method, the PPS method or the simple random sampling ratio estimation method should be adopted as much as possible in order to enhance the efficiency of cluster sampling. Furthemore, improving the overall estimation methodology is also a simple and economical way to enhance the efficiency of cluster sampling and therefore deserves more attention.

关 键 词：整群抽样 抽样效率 麦多方法 

分 类 号：F222.1[经济管理—国民经济]