概率抽样条件下样本代表性事后评估方法探讨

 样本代表性直接牵扯到统计数据质量和统计公布引起的民众反响，目前社会上不乏对政府统计数据的质疑之声，最终影响到政府统计机关的公信力，因此有必要重新审视现行样本代表性的研究。目前相关文献普遍强调不同抽样方式下的样本代表性的相对性内涵，确保样本的代表性仅从抽样方法和样本量两个维度入手，缺乏对既定抽样方法下样本代表性的事后评估体系的探索，以及多样本之间样本代表性优劣的比较方法研究。鉴于此，本文在结合人口普查数据基础上尝试从样本-总体整体分布和内部属性结构两个方面构建样本代表性事后评估的一整套指标和假设检验，进而找到一种多样本代表性比较的依据。最后对浦东新区2010年城镇居民收入调查样本进行了代表性检验的尝试。

Discussions on the Post-Evaluation of Sample's Representativeness in the Condition of Probability Sampling

The sample’s representativeness directly involves the quality of statistics and its public repressiveness,but nowadays sound of doubts arises frequently,which further influences the credibility of governmental statistical offices,so it’s necessary to reconsider the research on the sample’s representativeness.The current related literature has always been focused on its relativity between different sampling methods,to guarantee the sample’s representativeness generally set an emphasis on the sampling design and sample size,which lacks the research on post-evaluation of sample’s representativeness under the given sampling method and the comparable methods of multiple samples’ representativeness.This paper combining with the population census data to attempt to set up a series of indicators and hypothesis tests from the aspects of sample-population whole distribution and internal attribute structure in order to find a comparable basis between multiple samples.In the end,we make an attempt to evaluate the sample’s representativeness of the 2010 urban resident earnings survey in Pudong District.