Page 221 - Becoming Metric Wise
P. 221

212   Becoming Metric-Wise


          7.3.3 Independence
          The most significant disadvantage of the h-type index is that it is not
          independent in the sense of Bouyssou and Marchant (2011). If A and B
          represent sets of publications and citations, then strict independence for
          an indicator F means that if F(A) , F(B) and one adds to A and to B the
          same publications (with the same number of received citations) then still
          F(A) , F(B). This does not hold for the h-index. For instance:
                             A 5 4; 4; 4; 4; 0Š; then h AðÞ 5 4
                                 ½

                     B 5 5; 5; 5; 2; 1Š; then h BðÞ 5 3  and h BðÞ , hAðÞ
                        ½
             Adding two publications with 5 citations (one may even assume that
          A and B were coauthors on these two publications) gives:
               0                                                  0
                                          ½
             A A with added publicationsð  Þ 5 5; 5; 4; 4; 4; 4; 0Š; then h AðÞ 5 4
               0                                                  0
             B B with added publicationsð  Þ 5 5; 5; 5; 5; 5; 2; 1Š; then h BðÞ 5 5
                                          ½
                and consequently h AðÞ , hðBÞ:
             This example is based on a similar one presented in (Waltman & van
          Eck, 2012a).

          7.4 SIMPLE VARIATIONS ON THE H-INDEX
          Besides using different publication and citation windows one may restrict
          publications to those where the scientist is either first author or correspond-
          ing author (Hu et al., 2010), or take only the first author into account.
          Note that we do not support the use of these two methods as they might
          not give credit to some collaborators. The first of these methods might,
          however, be used in addition to the standard h-index. Another simple vari-
          ation is using complete-normalized fractional counting for citations (if an
          article is written with three authors, then each of them receives only one
          third of the citation credit), see Subsection 5.6.3.
             The original h-index is a natural number, leading to many scientists
          with the same value for their h-index. Two proposals have been made to
          refine this.

          7.4.1 The Rational h-index
          The rational variant of the h-index, denoted as h rat , was introduced by
          Ruane and Tol (2008) in the context of publications and citations (whole
          counting). It is defined as follows.
   216   217   218   219   220   221   222   223   224   225   226