The Researches On The Relationships Of Impact Factors, Cited Times And Downloading Times

The evaluation of journal papers is the important aspect of scientific research assessment. Public opinions are divergent about the evaluation of Journal papers, some people advocate that citation analysis and cited times are used to discuss; while others suggest using cited times and download frequency to evaluate; also some scholars evaluate the journals from one particular discipline or impact factors; some other scholars to explore from different aspects, such as: standard format, writing level, degree of innovation, practicability and so on. All of these can evaluate some particular characteristics of a paper, however, it is incomplete to evaluate the paper only using one indicator such as the cited times or impact factors.In order to give some guides for comprehensive evaluation of papers, according to the digital resources of chinese journals in CNKI database, the published papers of the branches of probability and statistics from the year 2010 to 2012 are selected as the research samples in this paper. We research the relationships of composite impact factors, comprehensive impact factors, cited times and downloading times, and explore the effectiveness and feasibility of these four indicators to evaluate the papers.The methods used in this paper are boxplot methods, comparative analysis, correlation analysis, factor analysis and clustering. The corresponding software for data processing and analyzing are SPSS21.0 and Excel2010.Firstly, boxplot methods are used to select the high-frequency cited and high frequency download paper; and then comparative analysis is used to observe the relationship between high download papers and high cited papers as well as the proportion of high download paper with cited number not more than 3; and then, relative analysis are used to discuss the relationship of download times and cited times. Lastly, factor analysis is used to extract the main factor and perform clustering analysis for high frequency papers.We come to the following conclusions by data analysis:(l) for the published high papers of the branches of probability and statistics in these three years, the downloaded times and cited times are positive correlation at the significant level of 0.01, after removing the singular value of each year, they are irrelevant or weak correlation at the same significant level. This conclusion tells us that download times and cited times can be the evaluation variables of journal papers at the same time.(2) for the journal papers, the only usage of impact factors, cited times or download times to evaluate the journal is incomplete and inappropriate. However, the comprehensive usage of them are feasible, what’s more, the evaluation results are more rational, more objective and more impartial.