统计数据准确性评估:方法分类及适用性分析

发布时间：2018-05-07 02:22

本文选题：统计数据准确性 + 评估方法　；参考：《统计研究》2009年01期

【摘要】：对于统计数据准确性评估方法的归纳分类,可以从辅助资料信息的来源及表现形态、包含目标特征真值关键信息的参照标准的构造方法、实际统计数值与参照标准之间的比较逻辑等方面加以考察。针对总量统计数据的准确性评估,主要在纵向时间维度上开展,根据统计指标与其相关联指标变动趋势的偏离程度加以评判;针对个体或分类统计数据的准确性评估则主要在横向空间维度上开展,对数据的统计分布形态加以检验,或者利用重复调查或随机实验方法对事先假定的误差参数进行估计。不同方法具有不同的适用条件,在实际应用中,应基于所掌握辅助资料的翔实程度来选择评估方法,并对导致评估方法发生误判的因素加以细致分析和排除,以确保评估结论的说服力和可信度。
[Abstract]:For the classification and induction of statistical data accuracy evaluation methods, we can construct the reference standard from the source and form of auxiliary information, including the key information of target feature truth value. The comparison logic between the actual statistical value and the reference standard is investigated. In view of the accuracy evaluation of total statistical data, it is mainly carried out in the longitudinal time dimension, which is judged according to the deviation degree of the change trend of statistical index and its associated index. The accuracy evaluation of individual or classified statistical data is mainly carried out on the horizontal dimension, the statistical distribution of the data is tested, or the error parameters are estimated by repeated surveys or random experiments. Different methods have different applicable conditions. In practical application, the evaluation methods should be selected based on the full extent of the auxiliary information available, and the factors leading to the misjudgment of the evaluation methods should be carefully analyzed and eliminated. To ensure the persuasiveness and credibility of the evaluation findings.
【作者单位】：厦门大学台湾研究院;中国人民大学统计学院;中国人民大学应用统计科学研究中心;
【基金】：国家自然科学基金项目“我国统计数据质量的评估系统研究”(项目编号:70671104)阶段性成果
【分类号】：C811

【参考文献】