研究与开发机构统计数据质量与异常点的对数正态分布检验与识别

一、引言　　统计数据来源于基层调查单位,对基层调查单位上报的原始数据进行检查和审核是提高统计数据质量的重要措施之一。这种检查和审核应包括以下两方面:(1)对基层调查单位上报的原始数据进行质量评估。通过评估对数据的质量有一个总体的认识和把握,特别是对数据的可靠性和准确性应有一个定量的说法。(2)对原始数据中的异常点进行识别和纠正。所谓异常点或称奇异点(ｏｕｔｌｉｅｒ),是对应于误差较大的观察数据,在这里主要是指社会经济和科技统计中的失实数据。由于种种技术上或非技术上原因形成的基层调查单位上报数据中的异常点,必然…

The Logarithm Normal Distribution Checking Method of Evaluating Quality and Identifying Outliers for the Statistical Data of the Research and Development Institutions

The evaluating the quality of original statistical data and identifying untruthful data from them are the key work and difficult point in statistical work. This paper has launched the evaluating and identifying by means of logarithm normal distribution according to the model presented by 3], and has taken the 1998 survey data from the research and development institutions as a example,. The result has authenticated the feasibility and usability of the method.