Abstract
The influence of scientific papers is measured by their citations. Although predicting the papers’ citation impact based on non-content factors has garnered extensive attention, the influence of such factors is rarely compared. In this article, we compare the influence of non-content factors on the citation counts of academic publications across three fields, i.e., math, computer science, and management. We consider different methods in this study, including three machine learning approaches, namely, XGBoost, Gradient Boosting Decision Tree, and Random Forest, along with statistical techniques such as linear regression and quantile analysis. Our findings reveal that no matter the field or analytical method applied, author prestige and the number of references consistently stand out as the most influential factors, while the breadth of categories covered by the paper has minimal impact. In mathematics, the first citation date and article length are almost equally important as author prestige, while the number of authors and the journal impact factor are crucial for computer science papers. In management, the number of collaborating countries is relatively influential with respect to the paper’s citations. The results of the quantile regression indicate that at higher quantile levels, the impact of author prestige and the number of authors on the papers’ citation impact are more pronounced across all three disciplines, while the journal impact factor and paper length have the greatest influence at low and medium quantile levels. Our findings indicate that the reliance of academic citations on author prestige and journal impact factors not only highlights the unequal distribution of resources within the current academic system but also further exacerbates citation inequality.
Published Version (
Free)
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call