A new approach to the metric of journals’ scientific prestige: The SJR indicator
Introduction
Citation analyses play an essential role in research evaluation systems, with their results being widely applied as complements to expert review.
The citedness of a scientific agent has for decades been regarded as an indicator of its scientific impact, and used to position it relative to other agents in the web of scholarly communications. In particular, various metrics based on citation counts have been developed to evaluate the impact of scholarly journals, one of which, the Impact Factor, has been extensively used for more than 40 years (Garfield, 2006).
However, recently a new research trend has emerged aimed at developing impact metrics that consider not only the raw number of citations received by a scientific agent, but also the importance or influence of the actors who issue those citations (Bergstrom, 2007, Bollen et al., 2006, Ma et al., 2008, Palacios-Huerta and Volij, 2004). These new metrics represent scientific impact as a function not of just the quantity of citations received but of a combination of the quantity and the quality.
The essential idea underlying the application of these arguments to the evaluation of scholarly journals is to assign weights to bibliographic citations based on the importance of the journals that issued them, so that citations issued by more important journals will be more valuable than those issued by less important ones. This “importance” will be computed recursively, i.e., the important journals will be those which in turn receive many citations from other important journals.
The first proposal in this sense in the field of Information Science was put forward by Pinski and Narin (1976), with a metric they called “Journal Influence”. Their proposed algorithm iterates the transfer of “prestige” from one journal to another until a steady-state solution is reached, whose values reflect the journals’ scientific influence. The “Journal Influence” indicator is a variant of the eigenvector centrality measure (Bonacich, 1987), with its calculation belonging to the group of eigenvector centrality methods in the domain of Network Theory. However, Pinsky and Narin's method presented problems that were essentially related to the topological structure of the citation network.
With the arrival of the PageRank algorithm (Page et al., 1998) developed by the creators of Google, one had a computational model that resolved the aforementioned structure-related problems (Brin and Page, 1998, Page, 2001). Inspired in the Perron–Frobenius theorem, this algorithm modifies the network's structure by redefining the meaning of the connections between the nodes that together conform the network's graph. In particular, it defines connections (citations) as the probability of going from one node to another, and, using a random-walker probabilistic model, transforms the citation network into a strongly connected graph, i.e., a network in which, given any two nodes, there is always some path to get from one to the other.
Applied to journal citation networks, this new model means that each connection between nodes (journals) represents the probability that a researcher, in documenting his or her research, goes from one journal to another by selecting a random reference in a research article of the citing journal. The values obtained at the end of the process represent a “random research walk” which starts from a random journal to end in another after following an infinite process of selecting random references in research articles. Also, to connect nodes (journals) between which there exist no paths established by means of citation relationships, a random jump factor is added to represent the probability that the researcher chooses a journal by means other than following the references of research articles.
The method also defines an iterative algorithm that starts from certain initial pre-established values, and computes values of centrality until a steady-state solution is reached. The importance (prestige) of the nodes is redistributed at each iteration in terms of their connections with other nodes. The general formula used in this process is:where the importance of node i in iteration k is set by the sum of the relative importance transferred by all the i-connected nodes. The amount of importance transferred by node j to node i is weighted by the strength of the connection between them, which is the fraction of references in node j in the year being considered that are to node i. The proportion of prestige that is transferred by means of the connections is modulated by means of the parameter λ which can take values in the range 0–1. The random jump factor, represented by the first term in the formula, is included to ensure convergence of the algorithm.
We introduce here a new indicator called SCImago Journal Rank (SJR) indicator, that indicates what can be denominated as journal's influence or prestige (Bollen et al., 2006), that belongs to this new family of indicators based on eigenvector centrality. The SJR indicator is a size-independent metric aimed at measuring the current “average prestige per paper” of journals for use in research evaluation processes. It has already been studied as a tool for evaluating the journals in the Scopus database (Guz & Rushchitsky, 2009), compared with the Thomson Scientific Impact Factor (Falagas, Kouranos, Arencibia-Jorge, & Karageorgopoulos, 2008), and shown to constitute a good alternative for journal evaluation (Leydesdorff, 2009). In studying both bibliometric and usage indicators, Bollen, van de Sompel, Hagberg, and Chute (2009) grouped the Impact Factor and the SCImago Journal Rank together, while clustering the Journal PageRank measure together with other “betweenness” centrality indicators. This was because the former are size-independent indicators rather than because they measure popularity as such.
In the following sections, we shall describe the methodological aspects of the development of the SJR indicator, and the results obtained with its implementation on Elsevier's Scopus database, for which the data were obtained from SCImago Journal & Country Rank website, an open access informetric directory with more than 17 000 research journals and other periodical publications (2009).
Section snippets
Data
We used Scopus as the data source for the development of the SJR indicator because it best represents the overall structure of world science at a global scale. Scopus is the world's largest scientific database if we look at the period 2000–2009. It covers most of the journals included in the Thomson Reuters Web of Science (WoS) and more (Leydesdorff et al., in press, Moya-Anegón et al., 2007). Also, despite its only relatively recent launch in 2004, there are already various studies of its
Method
The SJR indicator is computed over a journal citation network where the nodes represent the scholarly journals in the database and the directed connections among the nodes the citation relationships among such journals. In our approach in particular, a directed connection between two journals is a normalized value of the number of references that the transferring journal makes to the recipient journal. The normalization factor used is the total number of references of the transferring journal
Statistical characterization
We carried out a statistical characterization of the SJR indicator in order to contrast its capacity to depict what could be termed “average prestige per document” with the journals’ citedness per document. In the following paragraphs, we shall present comparisons of the rank distributions and scatterplots of the SJR indicator and the Journal Impact Factor, both overall for the entire database, and for some of the “subject areas” and “specific subject areas” de Scopus. We constructed an ad hoc
Conclusions
This study has presented the development of the SJR indicator, a new metric of the scientific influence of scholarly journals aimed at use in conventional processes of research evaluation.
Since it is constructed on the Scopus database, we believe it will best reflect the citation relationships among scientific sources. However, at the same time, it will be necessary to adapt the PageRank method of computation to the particularly complex and heterogeneously structured characteristics of a
Acknowledgements
This work was financed by the Junta de Extremadura—Consejería de Educación Ciencia & Tecnología and the Fondo Social Europeo as part of research project PRI06A200, and by the Plan Nacional de Investigación Científica, Desarrollo e Innovación Tecnológica 2008–2011 and the Fondo Europeo de Desarrollo Regional (FEDER) as part of research projects TIN2008-06514-C02-01 and TIN2008-06514-C02-02.
References (22)
- et al.
The anatomy of a large-scale hypertextual web search engine
Computer Networks
(1998) - et al.
Bringing PageRank to the citation analysis
Information Processing and Management
(2008) - et al.
Citation influence for journal aggregates of scientific publications: Theory, with application to the literature of physics
Information Processing and Management
(1976) Which h-index?—A comparison of WoS, Scopus and Google Scholar
Scientometrics
(2008)Eigenfactor: Measuring the value and prestige of scholarly journals
College & Research Libraries News
(2007)- et al.
Journal status
Scientometrics
(2006) - et al.
A principal component analysis of 39 scientific impact measures
PLoS ONE
(2009) Power and centrality: A family of measures
American Journal of Sociology
(1987)- et al.
Comparison of SCImago journal rank indicator with journal impact factor
The FASEB Journal
(2008) The history and meaning of the journal impact factor
JAMA-Journal of the American Medical Association
(2006)
Cited by (498)
Big data analytics capabilities: Patchwork or progress? A systematic review of the status quo and implications for future research
2023, Technological Forecasting and Social ChangeScience mapping for concrete composites as radiation shielding: A review
2023, Radiation Physics and ChemistryThe role of trust in family business stakeholder relationships: A systematic literature review
2023, Journal of Family Business StrategyArticle's scientific prestige: Measuring the impact of individual articles in the web of science
2023, Journal of InformetricsAcademic Production and Collaboration Among BRICS-Based Researchers: How Far Can the “De-Westernization” of Communication and Media Studies Go?
2024, Journalism and Mass Communication Quarterly