Research On The Relationship Between Social Media Attention And Citations Based On Mediation Effect Model

Will the attention raised by scholarly articles on social media affect their citation? It is an important issue worth studying.This paper is based on the Altmetrics data and citation data of 2612 papers published by PeerJ from 2013-2016.The mediation effect model is used to demonstrate the research hypothesis that the social media attention will positively refer to citation and there is an impact path from the social media attention to the views of the papers and then to the citation of the papers.The following findings were found in the study: First,the altmetric score,the number of HTML views or PDF downloads,and citations satisfy the Pareto rule,that is,about 20% of the papers have higher measurement index values,while 80% of the papers have lower measurement index values,and the overall distribution is power law.Second,the attention of social media has a significant impact on the number of visits to scholarly papers.The higher the attention of social media,the more significant the change in visits,but the significant impact is unsustainable and will decline as heats recedes on social media.The duration of the heat is about 7 days after the publication of the paper.Thirdly,the correlation analysis shows that the linear correlation between each pair of variables is statistically significant.The altmetric score is moderately related to the number of HTML views / PDF downloads and is moderately weak with citations.A comparative analysis of the control group showed that papers with high social media attention or HTML views or PDF downloads generally gained citations advantage than lower papers.The papers with high altmetric score are about 3.5 times higher than the lower papers on average.For high HTML views or PDF download papers,the figure is 5.4 and 5.2.Finally,the test results of the mediation effect indicate that path in the research hypothesis exists,and the effect produced by this path is 0.320,which HTML views accounts for 60% and PDF views accounts for 40%.