21 November, 2009

The Hirsch-bar index: leave your boss' name out of the authors list

I think you know what the Hirsch-index is: if you have n papers cited at least n times – then your h-index is n. For instance, if out of your 10 papers two were cited 4 times, and the rest three times, then your h-index is 3. The index was introduced by Jorge Hirsch in 2005, and became an accepted tool to rank scientists within about one year: community found it useful that a scientist's productivity may be described by a single number. However, the h-index does not take into account such important things as the number of co-authors and the self-citations.

Recently Jorge Hirsch himself wrote a preprint, where he proposed a new, ħ (hbar) - index, that is aimed to solve at least one of these problems, multiple co-athorship. Indeed, the "mass graves" (sorry), published by high-energy physicists and the institute policies to include everyone's name to any article, can unreasonably increase the Hirsch-index, with no extra work.

Some of the h-index terminology before describing the ħ-index. You have the h-index n, if n of your papers belong to the "h core". The paper belongs to the h-core if it was cited n times or more. Your co-authors will probably have other h cores, and a given paper may belong there or not.

ħ - index is defined in a similar way: you have the ħ - index m if m of your papers belong to the ħ-core. But, the paper belongs to the ħ-core if it was cited m times or more and it also belongs to the h-core of your co-authors. (Actually, the latter h should be also ħ, but since the index would be extremely difficult to compute in this case, Hirsch decided to relax the condition a bit)

In other words, only if the paper improves the h-index of all the authors, it will contribute to your ħ-index. Imagine that you have the h-index of 20, and you have a paper cited 25 times, co-authored by a student with a h-index of 5. Then, it will be included in your and the student's ħ-index, because it belongs to the h-cores of both of you. But, if you add a third author to this paper, say, a director of your institute with a h-index of 45, the paper will not count for any of you, because it doesn't belong to the director's h-core.

So, if you have 200 papers, cited 44 times each and co-authored by your director, who has the h-index of 45, none of them will count, and your ħ-index will be zero. Hirsch expects this to stimulate young scientists to work independently. However, he mentions, that the ħ-index will be definitely useless for postdocs, since they are used to include senior scientists in their papers.

I wonder whether it will be possible to eliminate self-citations at some point.


Anonymous said...

I think with that index you're lucky, if your boss is a young professor, who's gonna collect the citations o his life in the future. If you're a student of - say - Robert Axelrod, you will have to wait a whole lifetime, until your hbar improves, no matter, how good you are.

Unknown said...

Dear Anonymous,

Well, this is true, and Hirsch himself mentions that the hbar-index will not work for postdocs and even for some of assistant professors, not mentioning PhD students.

So, this index is expected to be used to rank scientists with tenure or tenure-track positions.


Anonymous said...

To anonymous above: not so lucky. If your boss is a young professor his letter of recommendation will not help you nearly as much as if he was Robert Axelrod writing for his lower hbar student.