论文部分内容阅读
以2012年JCR收录的数学期刊为例,采用偏度、峰度、JB检验、极大极小值比、离散系数、中位数均值比分析期刊评价指标的数据分布特点,并首次采用基尼系数分析期刊评价指标的内部差距,发现期刊评价指标普遍是右偏的,并且不服从正态分布;总被引频次、特征因子、即年指标的内部差距较大。期刊评价指标数据偏倚情况从好到坏的次序为:影响因子与5年影响因子>被引半衰期>论文影响分值>即年指标>特征因子>总被引频次。得出结论:指标数据偏倚会影响评价指标的数据标准化;指标数据偏倚会影响期刊一般水平的判断;指标数据右偏会导致期刊评价值偏低;最好选取数据偏倚情况相对较好的指标来评价期刊平均水平;数据偏倚对基于传统回归的计量研究影响较大。该结论有待进一步检验。
Taking JJJJJJJJJJJJJJJJJJJJ as an example, we analyzed the data distribution characteristics of journal evaluation indexes using skewness, kurtosis, JB test, maximum / minimum ratio, discrete coefficient and median mean ratio. The Gini coefficient Analyzing the internal gap of evaluation index of periodicals, we found that the evaluation indexes of periodicals are generally right-deviation and do not obey the normal distribution. The total internal frequency of citation frequency and characteristic factor, that is, the annual index is larger. The sequence of the evaluation of the journal evaluation index data from good to bad is as follows: influencing factor and five-year influencing factor> cited half-life> thesis impact score> annual index> characteristic factor> total number of cites. It is concluded that the bias of index data will affect the data standardization of evaluation index; the bias of index data will affect the general judgment of journal; the right bias of index data will lead to the low evaluation value of journal; it is better to select the index with relatively better data bias Evaluation of the average level of the journal; Data bias has a greater impact on the measurement based on traditional regression. This conclusion needs further examination.