Do we need the E-index in addition to the h-index and its variants?

Today, the h-index is a widely accepted indicator of scientific performance and has become a built-in feature of bibliographic databases such as the Web of Science (Thomson Reuters) and Scopus (Elsevier) (van Eck & Waltman, 2008). The wide acceptance is due mainly to the many advantages of the new index. These include its simplicity and intuitive meaning (Franceschini & Maisano, 2010), the combination of citation impact and publication numbers in one number (British Academy, 2007), and its robustness against small errors and single peaks (top-cited papers) in a publication list (Liu & Rousseau, 2009). In a recent issue of Physics Today, Gott (2010) criticized the h-index for measuring a scientist's output: It does not proportionally reward individuals for their most important papers and unfairly favors scientists in large collaborations. Both critical points are well-known to researchers and have led to some h-index variants that are supposed to not have these disadvantages (Bornmann, Mutz, Hug, & Daniel, in press). For example, the g-index proposed by Egghe (2006) places more weight on the citation performance of a set of papers than does the h-index. Against the backdrop of his critique on the h-index, Gott (2010) proposed " an E-index citation count, proportional to total output. In the E-index, the total count would equal 1/2 first-author citations + 1/2 normalized citations + last-name citations in abstracts + last-name citations in titles " (p. 12). Bornmann et al., in press) on the h-index published so far was to determine empirically the extent to which the development of the h-index variants does in fact result in additional information not provided by the h-index. Although the proposed variants may be conceptualized differently than the h-index theoretically or mathematically, in their empirical application, they may be highly correlated with the h-index. Are there h-index variants that make a nonredundant contribution to the h-index (i.e., that have a low correlation with the h-index)? For Burrell (2009), for instance, each of the h-index variants seems to be approximately proportional to time, and hence each is approximately proportional to each of the others. The results of the first meta-analysis that has been conducted on the h-index and its variants show, with an overall mean value between .8 and .9, that there is a high correlation between the h-index and 37 h-index variants (Bornmann et al., in press). However, not all h-index variants have a high correlation with the h-index. The …