Pretty much all systems of statistical performance measurement face a trade-off between meaningfulness and simplicity. Straightforward, easy-to-understand statistics usually don’t tell you very much because the process of simplification inevitably leaves out important aspects of reality; statistics which take complexity into account are usually clunky and difficult to explain to a lay audience.
So it is with bibliometrics. We can count scholarly publications, but what if someone is just publishing in obscure journals that no one reads? We can adjust publications by the importance of the journals in which articles are published (every journal has an “impact factor”), but that’s an indirect measure at best. We can count citations directly, but how does one adjust for the “worst-paper-ever” phenomenon (i.e., everyone cites a particular paper as being especially bad/wrong/useless)? While it is possible to correct for all these things, the resulting statistics get pretty tortuous.
However, one bibliometric statistic has come to the fore in recent years because of the way it combines simplicity and meaningfulness. That statistic is what’s known as the Hirsch-index (or H-index) after its creator, Jorge E. Hirsch. The H-index boils down scholarly productivity to a single number, which technically stated is “the largest number n, for which n of the researcher’s publications have been cited at least n times.” That sounds complicated, but in practice it’s pretty clear: if you have five papers with at least five citations each, your H-index will be five. If you have ten publications, one of which has a thousand citations and the others none, then your H-index is one.
The H-index’s genius lies in the way it combines both output and impact while limiting the influence of a small number of very successful articles, which may be unrepresentative of a researcher’s career. It is the best individual measure yet devised to sum up scholarly output.
Of course, H-index values vary significantly across disciplines because of differences in publication cultures. Biologists and physicists publish and cite a lot more frequently than, say, English professors, whose scholarly communications involve a lot fewer journal articles and citations. Simply comparing two scholars’ H-index values doesn’t tell you very much unless they are in the same discipline.
That is, of course, unless you have a system for accurately normalizing across disciplines. No one has yet done this because doing so would involve a tedious process of gathering tens of thousands of observations of individual scholars’ H-index results. But, of course, if you read last week’s Globe and Mail insert on research and development, you’ll know that we at HESA have recently completed exactly such a tedious process. And we’re about to publish some rather interesting results about it.