Scientific Evaluation Method Research Based On Ref

With constant development of scientific research, research assessment is highly focused by governments and all sectors of the community. Research evaluation promotes the development of science, while the development of science further provides a new challenge for research evaluation. However, both expert review and citation analysis have their own limitations in present research evaluation activities. Furthermore, due to different characteristics between disciplines, it is impossible that any single evaluation method is applied to make assessment results complete and effective.This dissertation is focused on research assessment and evaluation from two aspects: theory and practice.On the theatric side, first of all, this dissertation expounds the development process of science evaluation:namely, the construction in peer review, the rise in bibliometrics and the advent of citation analysis. It further analyzes the advantages and disadvantages of peer review and bibliometric indicators and points out the combination of qualitative and quantitative evaluation methods will be important to evaluation.Secondly, in view of characteristics of the combination of qualitative and quantitative evaluation methods and combining a new concept in research assessment—metrics-based REF, the dissertation studies the background, unit of assessment, panels, three modles of bibliometrics and main factors of evaluation framework of REF and rethinks the role of REF in research assessment.Finally, this dissertation focuses on the correlation of qualitative and quantitative evaluation methods and then constructs elevation framework of nature science and humanities and social sciences after the analysis of characteristic of nature science and humanities and social science.On the practice side, this dissertation integrates the results from bibliometric measures, inductive deductive method and literature survey. The author chooses1428papers in F1000:namely725ecological literatures and703psychological literatures and searches their related bibliometric indicators in Scopus and Web of Science. Then it compares F1000Ariticle Factor by expert review with other measures of paper’s impact by SPSS16.0in order to verify the association between expert assessment and bibliometric indicators. From this point of view of statistics, the result shows that there is significant positive correlation between expert assessment of importance and impact as measured by number of citations overall. However, the data analysis indicates that there are some exceptions that some papers that were highly rated by experts were not highly cited, and vice versa.The result of the correlations between F1000Ariticle Factor and bibliometric indicators shows that comparing to other indicaors, there is more relevant correlation between SJR or IF and expert assessment of performance.These results provide a strong basis for peer review.The conclusion of this dissertation is that sole reliance on bibliometric indicators or expert assessment would have made assessment result inappropriate.Bibliometric measures will not completely replace expert review, just supplement to expert assessment. Tools that link peer review of research quality and importance to bibliometric measures, such as citation analysis would be main stream of the field of research assessment and evaluation in the future. Baseed on research, the paper holds that it is necessary to build corresponding evaluation framework according to the different characteristics of the disciplines.Based on the new concept in research assessment—REF, the construction of evaluation framework of nature science and humanities and social science will provide reference for the related research on construction of research evaluation systerm framework in the future.