部分服务业抽样调查简介Sample survey on partial serviceshttp:/www.docin.com/sundae_meng主要内容主要内容（Outline）调查目的（Objectives of survey）调查范围（Industries of survey）调查方式（Way of survey）抽样精度（Precision of sampling）组织实施（Organization and implementation）探讨的问题（Questions)National Bureau of Statistics of China 调查目的调查目的（Objectives of survey）反映部分服务业的基本总量和发展速度，为国民经济核算提供基础数据。 Reflecting the total value and tempo of partial services, providing basic data for national accounts.National Bureau of Statistics of China调查范围调查范围（Industries of survey） 装卸搬运和其他运输服务业（Loading, unloading, portage and other transport services） 仓储业（Storage ） 计算机服务业（Computer services ） 软件业（Software industry ） 租赁业（Tenancy ） 商务服务业（Business service ） 科技交流和推广服务业（Services of science and technique intercommunion and generalization） 居民服务业（Resident services ） 其他服务业（Other services ） 体育（Sports activities ） 娱乐业 （Entertainment ）National Bureau of Statistics of China调查方式调查方式（Way of survey） 采用目录抽样方法进行调查，即直接从企业名录库中抽取样本企业。 Sampling directly from the BR(Bussiness Register) of the enterprises, with using the method of directory sampling. 抽样框（Sampling frame） 使用在2004年全国第一次经济普查资料基础上维护后的名录库，由各调查总队从各省级普查中心搜集并负责抽取样本。 Using the BR（Business Register）after the maintenance on the basis of China Economic Census in 2004. Provincial survey office of NBS sampling the businesses from the BR from provincial census center.National Bureau of Statistics of China调查方式调查方式（Way of survey） 样本量（Sample Size） 按满足省总量需要及全国分行业大类需要确定样本量，国家根据2004年全国第一次经济普查资料确定分行业样本量并分配到各省级，各调查总队在此基础上确定本地区样本量。 Determining the sample size with meeting the needs of total quantity by province and national quantity by industries. NBS determines the sample size by industries according to the data of China Economic Census in 2004, and allocates the samples to every province. On the basis of the allocated samples, provincial survey offices of NBS determine the sample size in their own areas.National Bureau of Statistics of China调查方式调查方式（Way of survey） 企业分层（Stratification） 在省内将企业子总体按行业大类分层，再按营业收入规模进一步分层。 Stratifying subpopulation by industry category in every province, and further stratifying by revenue. 样本抽取（Sampling） 利用永久随机数技术，按随机方法为每个企业赋予一个随机数，抽取 个最小永久随机数企业作为最终层的样本。 Giving a random number to every enterprise by the method of random using the technique of permanent random number, and sampling the enterprises with the minimum sample number as the samples in the final stratum. Here, means the sample number in the h stratum.National Bureau of Statistics of China 估计量（Estimations） （1）总量估计的公式为（the Equation for Total Estimation）： 为回答单位的最终权数， 是单位的指标值。 is the final weight of corresponding enterprise, is the value of corresponding survey variable 调查方式调查方式（Way of survey）National Bureau of Statistics of China调查方式调查方式（Way of survey） （2）子域总量估计的公式为（ The equation for total estimation in subfields ）： 为所要研究的子域(如行业等)， 为回答单位的最终权数， 是单位 的指标值。 is a certain subfield (such as industries), is the final weight of respondent enterprise, is the value of corresponding survey variable （3）方差估计（ Variance estimation ） 利用STATA软件进行方差估计 Using STATA software to estimate varianceNational Bureau of Statistics of China抽样精度抽样精度（Precision of sampling） 以省为总体控制总量指标抽样精度。具体要求是：在95的概率保证程度下，营业收入的最大相对误差控制在10以内。 Controlling the precision of sampling for total indexes by province. The specific requirement is that under the guarantee of 95% probability, the maximum relative error of revenue is controlled less than 10%.National Bureau of Statistics of China组织实施组织实施（Organization and implementation） 由服务业调查中心负责组织，各调查总队具体实施。 Organized by Service Survey Center of NBS, Implemented by provincial survey office of NBS.National Bureau of Statistics of China探讨的问题探讨的问题（Questions） 现在遇到的主要技术难题是多指标问题，即不能兼顾到各个指标的变异系数（CV）。我们是以营业收入作为辅助变量进行抽样设计分层的。这样，与营业收入相关性比较高的指标，其估计值的CV较小；但与营业收入相关程度比较低的指标，其估计值的CV就比较大。 The main technical problem encountered is the question from multiple indexes. That is, we cant take into account CV (coefficient of variance) of all indexes. We use revenue as the auxiliary variable in stratification of sampling design. Thus, the CV of the indexes highly relevant with revenue are smaller, but the CV of the indexes lowly relevant with revenue are larger.National Bureau of Statistics of China