Multiplicative versus fractional counting methods for co-authored publications : the case of the 500 universities in the Leiden ranking

This paper studies the assignment of responsibility to the participants in the case of co-authored scientific publications. In the conceptual part, we establish that the key shortcoming of the full counting method is its incompatibility with the use of additively decomposable citation impact indicators. In the empirical part of the paper, we study the consequences of adopting the address-line fractional or multiplicative counting method. For this purpose, we use a Web of Science dataset consisting of 3.6 million articles published in the 2005-2008 period, and classified into 5,119 clusters. Our research units are the 500 universities in the 2013 edition of the CWTS Leiden Ranking. Citation impact is measured using the Mean Normalized Citation Score, and the Top 10% indicators. The main findings are the following. Firstly, although a change of counting methods alters co-authorship and citation impact patterns, cardinal differences between co-authorship rates and between citation impact values are generally small. Nevertheless, such small differences generate considerable re-rankings between universities. Secondly, the universities that are more penalized by the adoption of a fractional rather than a multiplicative approach are those with a small co-authorship rate for the citation distribution as a whole, a large co-authorship rate in the upper tail of this distribution, a low citation impact performance, and a small number of solo publications.