Tri-dimensional Comparison Of The Correlation Between Paper Count And Times Cited Within Different Time Windows

As an important index for evaluating journals and papers,citation has been widely concerned and recognized,but there are many factors concerning for it,of which the correlation between paper length and total citation is quite controversial.The main reason is that the subject category involved,the level of journals published,and the type of literature of the papers are different.For that matter,this paper intends to explore the relationship between the length of papers and the total citations within different time windows through three dimensions of Web of Science(WOS)categories,journal quality(Journal Impact Factor Quartile represents journal quality)and document types.All the documents in 2018,2015,2012 and 2010 of 2 science disciplines(Materials Science,Biomaterials;Ophthalmology)and 2 social science disciplines(Information Science&Library Science;Sociology)in the WOS database were harvested to Excel,correspondingly.The field names of publication name,document type,core collection times cited count,page count,and year published in the obtained data retained.The page count represents the length of the paper.Only the documents whose document types are Article,Review,Editorial Material and Proceedings Paper were collected,and the Quartile(Q1,Q2,Q3 and Q4)was added according to the Journal Impact Factor Quartile in the Journal Citation Report by matching publication name.Then the Percentile in Subject Area(PSA)of the selected data from the In Cites database obtained.The citation analysis method was applied to compare the number and distribution of the times cited count and page count.The correlation analysis method was used to analyze the correlation between the page count and PSA,which was regarded as the general correlation between the times cited count and the length of paper,and then the correlation between the length of the paper and its citations was analyzed from three aspects of WOS categories,Quartiles and document types.Next,the harvested date were divided into 6 groups according to page counts:1 page to 5 pages,6-10 pages,11-15 pages,16-20 pages,21-30 pages,and more than 30 pages were group A,B,C,D,E,and F,respectively,and the correlation between the paper count and its citations was analyzed in the subgroups.The regression analysis method was used to measure the regression relationship between the paper count and times cited count,and the quantile regression method was applied to determine the degree of explanation of the paper count to the times cited count in different quantiles.In the results,it was found that the page count and times cited did not obey the normal distribution(all P<0.05).Correlation analysis between the length of paper and PSA found that the two were moderately negatively correlated(r=-0.301,P<0.001),indicating that the length of paper was moderately positively correlated with times cited.The R~2 of the linear regression equation of the two was 0.106,but as the length of the paper increased,the times cites did not increase without limit.Moreover,tri-dimensional correlation analysis showed that the citations in Information Science&Library Science,Materials Science,Biomaterials,Ophthalmology and Sociology were almost irrelevant to page counts(r=0.088,0.132,0.136 and 0.104,respectively,all P<0.01).The citations in Q1,Q2,Q3,and Q4 were also almost irrelevant to page count(r=0.124,0.137,0.199,and 0.154,respectively,all P<0.05).The Editorial Material paper count had a medium to high positive correlation with their citations(r=0.519,P<0.001),and the correlation coefficients of citation and page count in Article,Proceedings Paper and Review papers were 0.034(P=0.015),0.088(P=419)and 0.069(P=0.365),respectively.In different WOS categories and Quartiles,the linear goodness of fit of the two were all less than0.1,but the two variables in the Editorial Material paper had R~2=0.261,that is,the independent variable can explain 26.1%of the dependent variable.It showed that the correlation between the citation and pages was not affected by WOS categories and Quartiles,but was greatly affected by document types.Through the subgroup analysis,it was found that the shorter page groups,namely,the 1-5 pages group and the 6-10pages group papers had a relatively significant correlation between citation and page count(all P<0.05).Quantile regression showed that as the quantile of the dependent variable increased,the elasticity coefficient did not increase gradually,which suggested that even if the paper pages increased,the times cited cannot continue to increase.Moreover,the time window was an important regulation factor for the correlation between the page count and times cited count.In conclusion,the length of paper and times cites are generally in a moderately positive correlation.The relationship between the two is not affected by WOS categories and Quartiles,but is greatly affected by the document types.Although the two have a moderately positive relationship,as the paper count increases,the times cited count will not increase indefinitely.Therefore,it is not advisable,in order to increase the times cited count,to blindly increase or decrease the paper page count.More importantly,the author should be respected in deciding the manuscript page count based on explicitly describing the amount of information carried by the text itself.