Network Analysis of Academic Journals: Promoting Influential Research Through Collaboration

Abstract

Purpose: The collaboration patterns of scientists may affect the impact of their research. In particular, clustering among authors in scientific journals may stifle innovation. The purpose of this paper was to explore the relationship between author clustering and journal impact factor. Methods: Coauthor networks were generated for articles published in 2010-2015 in 31 journals within the fields of surgery and internal medicine. The overall degree of clustering within each journal was assessed by using network analysis techniques to calculate the average clustering coefficient (ACC). ACC values were compared between broad-interest and specialty-specific journals within surgery and internal medicine. Spearman’s correlation coefficient was calculated between each journal’s ACC and its impact factor, an established metric of a journal’s influence. Results: ACC was lowest in broad interest journals, like Science (0.014) and Nature (0.015), and clinical journals with a broad scope, like JAMA (0.025) and NEJM (0.026). In surgery and internal medicine, ACC increased as the journal became more specialized. There was a negative correlation between a journal’s ACC and its impact factor (Spearman’s rs=-0.49, p=0.005). Conclusions: Author clustering is higher in specialty-specific journals and is negatively correlated with a journal’s impact factor. Open collaboration networks may promote influential research.