Two Levels Quantitative Test Of Governmental Statistical Data Quality

The statistics is a weapon that knows the society. Along with the information of statistics becoming more important for the management of socialism market economy, social public gives more concerns to the statistical data quality, putting forward the higher request. In fact, the quality of the statistical data is a Life Line of statistics. But currently the research of the statistical data quality in our country is still weaker. For satisfying the social public in need of statistical data better in the process of developing of society economy, for bringing the statistics in our country further in line with the nations, It is necessary for our country to take a system-depth study for quantitative test of statistical data quality.First this text sets out from the statistical data and the connotation of the quality, introducing the concept of the statistical data quality of the overall system; Next, start with the research of the distribution regulation of statistical data, study the problem of accuracy test of statistical data. Using lognormal normal distribution test to carry on the quantitative test and the recognition to the statistical data quality and the unusual data which reflect research object scale size. And use relative fitting error to measure statistical data non- uniform error; Then introduce the method systematically of using the fuzzy comprehensive evaluation method to carry on the overall superior test of the government statistical data quality. Includes the establishment of step level appraisal target system, target weight determination, calculates the factor weight in various levels, uniform test of judgment matrix, and built up the final fuzzy comprehensive evaluation model of the government statistical data quality according to the above-mentioned standard; Finally selects the partial main social economy total quantity target from Chinese statistics yearbook 2003 to carry on the real diagnosis analysis: (1) Confirm these social economy total quantity targets using the description statistics and the K-S inspection method to obey the lognormal normal distribution. (2) According to the two levels of inspection methods which this article proposed to carry on accuracy and the overall superior test for these social economy total quantity targets. Thus achieves the goal of evaluate government statistical data quality comprehensively. Finally subtotal the merit and insufficient of this two levels of inspection methods. And proposed the direction how to improve diligently in the future.