统计数据准确性评估的误差效应分析方法

Error Effects Analysis Approach for Statistical Data Accuracy Evaluation

基于统计数据生产者角度开发有效而可靠的统计数据准确性评估方法,具有迫切的必要性。参照社会调查领域的计量误差效应评估技术,结合中国政府统计调查数据的结构特征,可以构造针对宏观统计数据的误差效应模型,从个体数据与总量数据两个层次、针对系统偏差与方差两类误差效应来定义统计数据的准确性。通过对该模型中若干重要参数的识别估计,既能实现传统的统计偏差评估,又可揭示因操作因素、制度因素、基层单元属性特征等对于统计数据准确性所造成的影响,从中发现重要的统计误差来源,从而推动统计方法制度进行有针对性地改革与完善。

It is necessary for statistical producer to develop some effectual and reliable evaluation methods of statistical data accuracy. Referring to the evaluation techniques of measurement error effects adopted in social survey programs and taking the structure characteristics of Chinese government statistics into account, an error effects model for statistical data has been constructed. Using the model, statistical data accuracy can be defined based on individual and total value two levels, and aiming at system bias and variance two types of error effects. After identification and estimation of parameters in the model, the systematic bias degree in statistical data can be evaluated, the effects of those factors such as statistical institution, field operation, and units＇ attribute on statistical data accuracy determined, significant statistical error sources detected, and statistical institution developed consequently.