大数据背景下小微企业抽样调查设计研究  

Research on the Sampling Design of Small and Micro Enterprises in the Context of Big Data

作　　者：万舒晨 Wan Shuchen(Department of Industrial Statistics,National Bureau of Statistics of China)

机构地区：[1]国家统计局工业统计司

出　　处：《调研世界》2024年第1期78-88,共11页The World of Survey and Research

基　　金：国家社科基金青年项目“大数据背景下基于单位名录库的统计调查问题研究”(17CTJ009)资助。

摘　　要：在大数据背景下,本文充分利用数据来源渠道丰富、网络调查较为便利等特点,从多重抽样框设计、利用辅助信息提升抽样推断精度、非概率抽样调查设计等角度入手,对小微企业抽样调查的研究思路进行了梳理,研究并总结了一些可以使用的办法。对于多重抽样框设计,本文在比较多重名录抽样框设计、多重名录抽样框和区域抽样框组合设计研究思路的基础上,得到了在一定调查费用限制下可优先选用重叠划分部分筛选法设计的结论;对于辅助信息的利用,本文总结了样本轮换有关模型,比较了多种设计的结果,得到了可优先选用轮换及保留样本单位均带辅助信息模型的结论;对于非概率抽样网络调查设计,本文研究了三种调查形式以及准随机、超总体模型、准随机与超总体模型结合三种抽样推断的思路,对于时效性要求较高的专项调查,可在候选者数据库抽取样本单位开展调查并进行非概率抽样推断。本文为大数据背景下小微企业抽样设计提供了科学依据,可结合实践进一步加强应用,提升小微企业抽样调查的效率。In the context of big data,this paper fully utilizes the characteristics of multiple data sources and convenient online surveys,starting with multiple frame design,using auxiliary information to improve sampling accuracy,and non-probability sampling design,to summarize the research ideas for sampling surveys of small and micro enterprises and some available methods.For the multiple frame design,this paper studies the design of multiple listed frames,the combination of multiple listed frames and an area frame,to draw a conclusion that the screening method is preferred for design under survey cost constraints.For the using of auxiliary information,this paper summarizes relevant sample rotation models,compares the results of various model designs,and draws a conclusion that the model of both rotation and retention sample units with auxiliary information can be selected preferentially.For the non-probability sampling design of online surveys,this paper studies three types of survey methods and the ideas of quasi-random,superpopulation model and the combination of quasi-random and superpopulation model for sampling inference.For special surveys with high timeliness requirements,sampling units can be selected from the candidate database to conduct surveys and non-probability sampling inference.This paper provides a scientific basis for the sampling design of small and micro enterprises in the context of big data,which can be further applied in combination with practice to improve the efficiency of sampling surveys for small and micro enterprises.

关 键 词：小微企业 大数据 多重抽样框 辅助信息 非概率抽样 

分 类 号：C811[社会学—统计学]