Refining the H-Index

Braun and colleagues recently examined the utility of the h-index (the number h of papers, each of which is cited at least h times) for assessing the impact of journals, and drew attention to some differences between the top 21 journals ranked according to the h-index and the journal impact factor. Their 4-year window, however, is inadequate. Data from the Web of Science suggest that the h-index for journals increases more-or-less linearly with time until it plateaus at about the twice the cited half-life, so it may be possible to base comparisons on a standard window (e.g., 3 years to be comparable with the journal impact factor), standardized by multiplying by the cited half-life divided by the width of the window (e.g., 3 years). Such an adjustment to the top 21 journals in Braun's table would promote the Journal of the American Chemical Society several places (from rank 20 to rank 6, if no external candidates are considered) and demote Nature Medicine (because of its youth, it has a short cited half-life). The use of a standard interval, without regard for the publication frequency of the journal or the nature of the discipline, introduces bias into both the journal impact factor and the h-index when applied to journals.