Does the h index for assessing single publications really work? A case study on papers published in chemistry

Schubert (Scientometrics, 78:559–565, 2009) showed that “a Hirsch-type index can be used for assessing single highly cited publications by calculating the h index of the set of papers citing the work in question” (p. 559). To demonstrate that this single publication h index is a useful yardstick to compare the quality of different publications; the index should be strongly related to the assessment by peers. In a comprehensive research project we investigated the peer review process of the Angewandte Chemie International Edition. The data set contains manuscripts reviewed in the year 2000 and accepted by the journal or rejected but published elsewhere. Single publication h index values were calculated for a total of 1,814 manuscripts. The results show a correlation in the expected direction between peer assessments and single publication h index values: After publication, manuscripts with positive ratings by the journal’s reviewers show on average higher h index values than manuscripts with negative ratings by reviewers (and later published elsewhere). However, our findings do not support Schubert’s (2009) assumption that the additional dimension of indirect citation influence contributes to a more refined picture of the most cited papers.