基于平衡样本的校准估计研究--以规模以下工业企业测算为例

Calibration Estimation Method Based on Balanced Samples--Taking the Measurement of Industrial Survey under the Designated Size as An Example

为解决规模以下工业企业调查中存在的样本代表性不足的问题,提出基于平衡样本的校准估计方法,并得出相应的估计量和估计量方差。该方法在抽样设计阶段采用了平衡抽样设计,在估计阶段采用了校准估计方法,较大限度地使用了辅助信息;通过数据分析得出基于平衡样本的校准估计方法要优于基于平衡抽样的HT估计方法。同时,为满足平衡变量间线性无关的假定,提出使用主成分分析、切片逆回归和切片平均方差估计三种方法对相关的平衡变量进行处理的思路。该方法对我国规模以下工业企业调查的完善具有理论与实践的双重意义,可适当的推广至我国政府统计的其他调查中。

In order to solve the problem of insufficient representativeness of samples in industrial enterprises under the designated size,this paper proposes a calitjration estimation method based on balanced samples,and obtains the corresponding estimator and estimator variance.This method adopts balanced sampling design in sampling design stage,calibration estimation method in estimation stage,and using auxiliary information to a great extent.Through data analysis,it is concluded that the calibration estimation method based on balanced samples is better than the HT estimation.At the same time,in order to satisfy the assumption of linearly independent between balanced variables,three methods of principal component analysis,sliced inverse regression and sliced average variance estimation are proposed to deal with the related balanced variables.This method has both theoretical and practical significance for the improvement the survey of industrial under the designated size of China,and can be appropriately extended to other surveys of government statistics.