The Author-Level Metrics Study: An Analysis of the Traditional and Alternative Metrics of Scholarly Impact for Neurosurgical Authors

Background and objective There is a paucity of information regarding the concordance of traditional metrics across publicly searchable databases and about the correlation between alternative and traditional metrics for neurosurgical authors. In this study, we aimed to assess the congruence between traditional metrics reported across Google Scholar (GS), Scopus (Sc), and ResearchGate (RG). We also aimed to establish the mathematical correlation between traditional metrics and alternative metrics provided by ResearchGate. Methods Author names listed on papers published in the Journal of Neurosurgery (JNS) in 2019 were collated. Traditional metrics [number of publications (NP), number of citations (NC), and author H-indices (AHi)] and alternative metrics (RG score, Research Interest score, etc. from RG and the GS i10-index) were also collected from publicly searchable author profiles. The concordance between the traditional metrics across the three databases was assessed using the intraclass correlation coefficient and Bland-Altman (BA) plots. The mathematical relation between the traditional and alternative metrics was analyzed. Results The AHi showed excellent agreement across the three databases studied. The level of agreement for NP and NC was good at lower median counts. At higher median counts, we found an increase in disagreement, especially for NP. The RG score, number of followers on RG, and Research Interest score independently predicted NC and AHi with a reasonable degree of accuracy. Conclusions A composite author-level matrix with AHi, RG score, Research Interest score, and the number of RG followers could be used to generate an "Impact Matrix" to describe the scholarly and real-world impact of a clinician’s work.