On the value of author indices

Abstract

In the November 2010 issue of PHYSICS TODAY (page 12), J. Richard Gott III criticizes the h-index as an inaccurate measure of the impact of a given author’s work; he recommends its replacement by an index he calls the E-index in honor of Albert Einstein. The h-index was proposed in 2005 by Jorge Hirsch,11. J. E. Hirsch, Proc. Natl. Acad. Sci. USA 102, 16569 (2005). in a paper that has been cited about 660 times in five years, a phenomenal impact for work on bibliometrics. The main reason for the large impact is the h-index’s simplicity and transparency and the ease of obtaining the index using the Web of Science. At least 37 variants of Hirsch’s h-index have been proposed since, and an E-index already exists.22. C. -T. Zhang, PLoS ONE 4(5), e5429 (2009). Some of Gott’s objections, and ways to take them into account, have been dealt with in the literature. His first objection is that the h-index does not reward an individual for his or her most important paper. Those of us who use the h-index profusely are aware of that concern and simply correct it by looking at the number of citations for the individual’s two or three most cited papers.Gott also objects to the low h-index of scientific giants like Einstein. That also has been discussed in the literature and attributed to the different publication and citation culture—in particular, the low average number of citations per paper—in Einstein’s day.33. W. Marx, L. Bornmann, M. Cardona, J. Am. Soc. Inf. Sci. Tech. 61, 2061 (2010). A normalization method to correct that problem through time adjustment is presented in reference [3]3. W. Marx, L. Bornmann, M. Cardona, J. Am. Soc. Inf. Sci. Tech. 61, 2061 (2010). . Once that procedure is applied, Einstein reaches an h-index of 139, higher than any of those mentioned by Gott.Another objection concerns informal citations—that is, those only mentioning names or initials instead of giving complete references—which Gott calls eponymous or name citations. Again, that has already been discussed at length.44. W. Marx, M. Cardona, Scientometrics 80, 1 (2009). The most conspicuous case may be that of C. V. Raman, who has an h-index of 17 with 2500 formal citations; 150 000 informal citations according to INSPEC, the Information Services for the Physics and Engineering Communities database; and 1 million informal citations according to Google Scholar.Combining formal and informal citations into one index is somewhat misleading because they are based on different metrics with different limitations. For example, abstracts from papers prior to 1991 are not available in the Web of Science. Furthermore, the weighting factors Gott used for calculating his E-index seem to be highly arbitrary, which makes his index opaque, especially when compared with the h-index. Finally, there is extensive literature discussing the relative share of first authors and coauthors in bibliometrical indicators.55. L. Egghe, J. Am. Soc. Inf. Sci. Tech. 59, 1608 (2008). REFERENCESSection:ChooseTop of pageREFERENCES <<CITING ARTICLES1. J. E. Hirsch, Proc. Natl. Acad. Sci. USA 102, 16569 (2005). Google ScholarCrossref, ISI2. C. -T. Zhang, PLoS ONE 4(5), e5429 (2009). Google ScholarCrossref3. W. Marx, L. Bornmann, M. Cardona, J. Am. Soc. Inf. Sci. Tech. 61, 2061 (2010). Google ScholarCrossref4. W. Marx, M. Cardona, Scientometrics 80, 1 (2009). Google ScholarCrossref5. L. Egghe, J. Am. Soc. Inf. Sci. Tech. 59, 1608 (2008). Google ScholarCrossref© 2011 American Institute of Physics.