Study On The Evaluation Methods Of Official Statistical Data Quality And Its Application

At this stage, the statistical data quality is of particular importance, which not only relates to both the effectiveness of national macro-control and the vital interests of economic entities in the various markets, but also relates to the statistical department’s and even the whole government’s credibility. At the meantime, generally speaking, our official statistics is not with high quality, which is far from the level of the developed countries;on the other hand, the study on China’s statistical data quality is relatively weak, in particular, the study on statistical data quality about our government’s actual conditions and evaluation methods are not enough. Thus, the systematic research on the evaluation methods of official statistical data quality, and making coordinated application research is of important theoretical significance.Firstly, the paper analyzes the basic theory problems relating to the evaluation of official statistical data quality. Quality of statistical data is the total features and characteristics of statistics which meet the explicit or implicit needs, including its accuracy, relevance, timeliness, consistency, accessibility and interpretability and other basic features. And factors affecting the quality of statistical data can be summarized as two inter-associated aspects, namely, statistical system design and statistical practical operation. Accordingly, based on the analysis, the thesis put forward the point the three parts on the statistical data quality evaluation at present Chinese government:the evaluation of statistical system design quality, the evaluation of statistical data accuracy, and the comprehensive evaluation of the statistical data quality.Secondly, according to the one-dimensional and multi-dimensional evaluation methods, the evaluation methods of statistical data quality are systematically cleaned up and summarized, the advantages and disadvantages of various evaluation methods and applicability are analyzed in detail, and the selection principal on the statistical data quality evaluation methods of practical applications were given, basing on which to form the method library of the statistical data quality assessment. Evaluation methods of one-dimension include logic test methods, econometric model analysis methods, accounting data revaluation methods, statistical distribution test methods and assessing errors in surveys methods, while multi-dimensional evaluation methods include mainly the following three basic categories, namely ROSC-DM method, indice evaluation methods and data quality comprehensive evaluation method. Thirdly, the paper chooses or re-designs appropriate assessment methods according to the current methods on statistical data quality evaluation, systematically evaluates the quality of our official statistical data from three aspects. The first thing is to design the assessment method on design method of GDP statistics in its statistical system with the idea of ROSC-DM method and comprehensive evluation method, analyzes the design quality of our government statistical system taking the GDP statistics as an example, and conclude that the design method of our GDP statistical system just complies with relevant international statistical norm in general. The second thing is to choose the econometric model analysis method as the basic evluation tool, and analyzes the accurancy of our GDP statistical data. It shows that the GDP data in2009and2010reflect our current economic situation and is accurate and credible in case the GDP data is accurate enough during1978to2008; our increasing data of GDP accounted by the past price during1985to2009is believe to bs very reliable as a whole, however, it lacks of accurancy in some years. For example, the GDP index in1990account by the price in1978has some problems. This paper shows the reliability is about288.7%, which is7%more than the data published by the authority; it means the chain growth speed in1990and1991has also some problems, whose reliasbility respectly are6.4%and6.6%, the former is2.6%more than the data published by the government and the latter is2.5%less than the official data. The third thing is to comprehensively evaluate the quality of our official statistical data with the fuzzy comprehensive evaluation method and weighted geometric average method from the data users’ aspect. Based on the index system of quaternary structure, on the one hand, the paper collects the basic evluation data through customer support satisfaction survey; on the other hand, it obtains the index weight data with the expert consultation. It comes to conclusion the two results are consistent, which means the quality of our official statistical data of perception by the data users is just in general level, the product quality of statistic data, its released quality and environment quality of statistic sysytem are also in general level, however, the product quality of statistic data is better than the other two. The consistancy of two conclusion shows both methods can be confirmed, therefore, to some extent, these two evaluation methods are both proper comprehensive assessment methods to evaluate the quality of statistical data.Finally, according to the conclusion of the theory and empirical research, and combining with the other realities of Chinese official statistics, based on the direction of promoting the international connection of Chinese official statistics, promoting the reform of the official statistics and establishing the quality evaluation system of official statistical data, it puts forward relevant measures to improve the official statisticial data quality of China.