One More Author-Level Metric: the g-index

The list seems endless, but I thought I would include one more author-level metric before moving on. I wanted to talk a little bit about the g-index and how it compares with the h-index. Now there is nothing wrong with the h-index, as long as it is interpreted properly. In general, the h-index correlates well with other measures of professional success.(1) However, because it weights publications over an entire career, it tends to favour researchers who are fairly consistent in their citation counts and are later in their careers, while at the same time it perhaps allows too much influence for early publications which are no longer representative of a researcher’s professional growth. None of this is a big deal – just something to keep in mind.

The g-index addresses some of limitations of the h-index by giving more weight to publications with higher citation rates – publications which are generally more impactful and so likely representative of current productivity. Thus, whereas the h-index is determined based on the individual citation counts of the top articles, the g-index is based on the average citation count of the more cited articles. Therefore, if, for example, the top 10 articles, by citation count, have an average score of 10, then that is the author’s g-index, even if one or two of the top 10 articles had quite low scores. A high scoring top article can compensate and pull up what would otherwise be a lower h-index.

One consequence of this is that g-indices are always at least equal to and are usually higher than h-indices. Well, that certainly makes authors feel better. The g-index also produces a wider spectrum of scores for the same set of researchers, and so it provides better discrimination – although it is not quite clear whether there is much difference, for example, between someone with a g-index of 12 and someone with a g-index of 13 – especially when these metrics are only meant to supplement qualitative assessment.(2)

A common problem with all metrics is that they are metrics. They are one-dimensional numerical measures which scarcely capture the subtleties of a research career. Next week, perhaps we should take a look at another, and to me more useful, method of capturing a researchers’ publication profile – the beam plot. I think of this as a graphical tool rather than a metric because it allows a lot of scaling to suit the field and genre of the individual researcher. Tune in next week.


  1. MacDermid JC, Fung EH, Law M. Bibliometric analyses of physical and occupational therapy faculty across Canada indicate productivity and impact of rehabilitation research. Physiotherapy Canada 2015;67(1):76-84.
  2. Thompson DF, Callen ER, Nahata MC. New indices in scholarship assessment. Am J Pharm Educ 2009;73(6):111.

Leave a Comment

Your email address will not be published. Required fields are marked *