科技评价指标值与评价属性背离及修正研究——基于多属性评价视角

Purpose/significance] This paper analyzes the deep-level reasons for the abnormality in the index distinction degree and the bias of data distribution in the science and technology evaluation and believes that this is a deviation between the evaluation index value and the evaluation attribute essentially. That is, the evaluation index value can not reflect the essence of the evaluation attribute well.[Method/process] This paper proposes a new method to reduce the deviation between evaluation index value and evaluation attribute which is logarithmic median standardization, and takes JCR2016 mathematics journal for an example to conduct an empirical analysis.[Result/conclusion] The results show that citation indexs are more likely to show deviations between evaluation index value and evaluation attribute. The deviation between evaluation index value and evaluation attribute can be determined from multiple perspectives, such as index connotation analysis, pass rate, dispersion coefficient, median maximum ratio, concentration index HHI, etc. The use of logarithmic median standardization can greatly reduce the deviation of evaluation index value and evaluation attribute.