Because I Need More Stress

Steinn reports a new metric for research productivity that some people are using: the "H-number":

The H-score, takes all your papers, ranked by citation count; then you take the largest "k" such that the kth ranked paper has at least k citations.

So, you start off with a H-score of zero.

If your 5th highest cited paper has 5 citations but your 6th highest cited paper has 4 citations then your H=5.

If your 10th highest cited paper has 11 citations, but your 11th highest cited paper has 9 citations, then your H=10.

And so on. High H is better.

Yeah, that's just what we need, another quasi-objective number for junior faculty to obsess over...

My score is 7, by the way, which not coincidentally is the total number of papers I've written that have non-zero citation figures recorded by the Harvard-Smithsonian abstract service. That's out of a total of 12 articles published in paper journals or proceedings-- 13 if you count the forthcoming PRA (which is moderately likely to take the H-score to 8 in a year or two), so I think that's not too shabby.

I would guess that these figures are probably significantly lower for experimentalists than theorists, as it takes longer to perform an experiment than to model one. I'm too busy this morning to try to figure out what the typical figures for an experimentalist would be, though.

More like this

that is in fact the exact context in which the H-score arose in conversation;
source of stress for junior faculty...
An interesting number of, and sort of, senior faculty have surprisingly encyclopedic knowledge of various people's H-score. I get the impression it circulated very rapidly to those on committees which seek out "objective quantitative" measures, and that a lot of H and dH/dt scores were calculated for a lot of unsuspecting people.

I got an unsolicited postdoc application from someone whose sole letter of recommendation focused almost entirely on the fact that the guy's H score was high for his age. The letter writer honestly seemed to think that this was sufficient to make him a strong candidate for a position.

I am not going to calculate my H-number, because it will make me weep.

I do not need to see some sort of quantification to know that at the moment my publication record is not what it should be. That's what happens when you switch research fields half-way through your pre-tenure career.

If somebody else calculates my H-number and tells it to me, I will [insert suitable threat here].

-Rob

My score is 7, by the way, which not coincidentally is the total number of papers I've written...

I think that sums up the experimental/theorist difference in H-scores; it takes a lot longer for an experimentalist to produce a paper, but once it's produced it'll almost certainly be cited a lot, whereas a theorist can churn them out but perhaps because of that they can be much more easily ignored.

My score is also 7, but only because of a lot of self- and close-collaborator- citing. Actually, depending on what you count, it can be up to nine -- the data from adsabs, ISI/WebOfScience, and scholar.google.com all give slightly different numbers.

Hmm. A quick peek at the arXiv reveals that my H score is "1", as one of my papers has 1 citation, and the other has none.

Let's hope nobody in my administration learns about this new metric, shall we?

I get H = 2 for me, which I think stands for ha-ha.

By Tom Renbarger (not verified) on 27 Oct 2006 #permalink

Bill -- I'm completely with you on not wanting to rely on some proprietary expensive service. (From the comments you linked to.) Hell, I'm one of those free culture nutcases, so you coiuld have predicted that, but in any event it's always a little scary to have some sort of performance metric controlled by a single proprietary interest. (And, yes, it may well be that I'm trying to make a snarky veiled reference to the College Board here.)

The thing about Google Scholar is that it's a little haphazard. I'd be hesitant to trust something like the h-index based on its data, beause some things are missing, some things are multiple cited.

At least in my field, I could get away with just using adsabs.harvard.edu.

And, sigh, yes, I just went and looked up my H-index there, which is 15, although who knows if adsabs is really getting all the citations. (On the most-cited paper, no, becuase that paper has been cited thousands of times by a million disciplines, but on the 15th, probably yes.)

If, on the other hand, I ONLY count first-author papers, my H-index is 4, which makes me look kinda pathetic, and is probably closer to reality....

Are you supposed to use only first-author papers, or all papers on which you appear as an author? Obviously, it makes a big difference....

-Rob

"Are you supposed to use only first-author papers, or all papers on which you appear as an author? Obviously, it makes a big difference...."

I assumed first-author only. I might actually threaten double figures if I went with all papers I am co-author on.

By Tom Renbarger (not verified) on 27 Oct 2006 #permalink

Isn't authorship kind of funny?

I am third on a paper that I did a ton of work for, over many years. That paper now has like 70 citations. I do not deserve as much credit as the first author, but I certainly deserve some.

But there are paper out there I was included on by virtue of breathing at the right time. Maybe it could be waited by placement? Just what we need! A new statistic!

I am surprised about how much weight is placed on these statistics. Given that, say, people who build instruments can be incredibly productive without every publishing, while big N-Body shops are very good at cranking out useful papers, a consistent metric seems dubious. This would only be useful in ranking people across a very narrow range.

By Brad Holden (not verified) on 27 Oct 2006 #permalink

I am third on a paper that I did a ton of work for, over many years. That paper now has like 70 citations. I do not deserve as much credit as the first author, but I certainly deserve some.

Yep. That's part of the problem with any bean-county statistic.

I was part of the SCP. Our competitors were the HZT. Same exact science, and in principle the same field. However, very different publication philosopies. When the accelerating Universe was discovered, the HZT had a post-doc as first author; the SCP had the team leader. Also, the HZT published a larger number of smaller papers.

The famous "Perlmutter et al. 1999" Astrophysical Journal paper (that has thousands of citations, and at one point was the most-cited paper (over some time period) in the particle physics literature) is one that I did a huge amount of work on. I'm one of some 30 or 50 or some such authors, but there were a number of us who did quite a bit of work and made huge contributions to that paper. (Indeed, if you look at the author list carefully, you will see that it's in three alphebetical tiers; first author, second group, third group.) Under any "first author" counting, I get no credit for this paper, but under any other counting, I probably get more credit for some other than I really ought to.

So....

Thought should be applied. Of course, that's effort, so never mind.

-Rob

Are you supposed to use only first-author papers, or all papers on which you appear as an author?

Baal's bollocks, man! It never occurred to me NOT to include absolutely everything; I shudder to think what my miserable score would drop to if I only counted first-author papers.

(And there again, my latest pub has me listed as first author, but it's a review and really my boss did the vast bulk of the work; so I'd be counting that, but really don't deserve much credit for it.)