Ballester, C. and A. Rosa
Keywords: journal,impact,eigenfactor,impact factor,pagerank,measurement,influence,centrality,self-citations,outerfactor
Abstract: In this paper, we use research chains across the citation graph as the basis for journal impact analysis. While some existing measures take into account research chains that end in a given journal, we calculate the proportion of research chains that include a journal, obtaining a new index of journal impact, Outerfactor, that is directly related to Pagerank (Brin and Page, 1998), Eigenfactor (Bergstrom, 2007) and the Invariant Method (Pinsky and Narin, 1976). In this way, the Outerfactor score obtained by each journal is independent on its own citation pattern and its article share. To our knowledge, this is the first measure that satisfies these invariance properties whilst accounting for both direct and indirect impact. Based on research chains that connect two journals, we also construct new measures for analyzing cross-impact. This cross-impact analysis results in a two-fold view of Outerfactor in terms of a journal’s influence (impact) on other journals, or a journal’s contribution to all journals’ impact scores. Finally, we provide an illustration with 60 economics journals, showing how Outerfactor performs compared to other measures: apart from its cardinal invariance, Outerfactor behaves more robustly to ordinal manipulation than other eigenvector-based measures.