Publications Repository - Gdańsk University of Technology

In this paper, we investigate the properties of the science-wide author databases’ stand- ardised citation indicators produced since 2015 by a team of researchers led by John P. A. Ioannidis. Based on data from Scopus, the researchers published close to 200,000 of the most-cited authors across all scientific fields and ranked them using a composite indica- tor that includes six citation metrics (total citations; Hirsch h-index; coauthorship-adjusted Schreiber hm-index; the number of citations to papers as a single author; the number of citations to papers as a single or first author; and the number of citations to papers as a sin- gle, first, or last author). We look at the quality of these new rankings to suggest improve- ments. Specifically, we try to correct for correlation among variables and the deviation between the purported weights of the variables, as declared by the developers, and the effective weights based on the theory of global sensitivity analysis. We aim to ascertain if our modified measure is an improvement over the existing one by addressing the issue of redundant data due to correlated variables. Lacking a ground truth based on a ‘true’ ranking possibly supported by independent data, we content ourselves for the purpose of the present analysis with benchmarking against a best practice. Given that most practition- ers have reservations against linear aggregation approaches (Munda, Social multi-criteria evaluation for a sustainable economy, Springer, Berlin, 2008; Balinski & Laraki, Major- ity judgment: Measuring, ranking, and electing, Mass, 2011; Arrow, Social choice and individual values, Martino Fine Books, Eastford, 2012 [1951]), while those based on the method of Condorcet are considered by many as a good practice (Munda, Social multi-cri- teria evaluation for a sustainable economy, Springer, Berlin, 2008; Roy, Multicriteria meth- odology for decision aiding, Springer, Cham, 1996), we entrust of defence of our modified ranking on a pairwise comparison between linear aggregation and Condorcet

Authors

• dr Marta Kuc-Czarnecka link open in new tab ,
• prof Andrea Saltelli,