A quick search on Google Scholar found that all ten of my most cited papers have multiple authors. Looking up the top ten most cited papers from some of the other tenured faculty in our statistics department: Shaw-Hwa Lo (9/10 have multiple authors), Zhiliang Ying (9/10), Daniel Rabinowitz (9/10), Ioannis Karatzas (8/10), Victor de la Pena (7/10). (I tried to look up Chris Heyde also, but Google Scholar kept coming up with articles referring to him rather than articles by him.) Victor and Ioannis are probabilists–their work is closer to pure math so perhaps it makes sense that their single-authored papers are (relatively) more prominent.
Anyway, I think it’s an important point, since it’s easy to undervalue multiple-authored work by diluting the credit among all the authors.
Dear Prof. Gelman,
From the young student point of view, the one aiming, perhaps, to stay in the academia once completed his PhD in Statistics, and according to your experience, do you think it is better to (try to) publish more multi-authored (multi-disciplinary, applied) papers (on more important journals ) or fewer single authored (most likely on less important journals) ?
with reference and many compliments for the nice idea of having a blog,
daniel
I'd also be interested in each author's overall proportion of multiple-author papers. The numbers you cite would be more impressive if, say, only 40% of each author's total papers have multiple authors.
Maybe the base rate of multiple author papers accounts for the difference between the probabilists and the others?
Dani,
I'm probably the wrong person to give advice on this. In the early 1990s, a colleague in my department heard that I was writing a book and he suggested I stop because it would be bad for my tenure prospects. I replied that I'd rather write the book than get tenure. More seriously, I think that an academic jobs at a good university is a great opportunity to work on whatever you think is important.
Ed,
I see what you're saying, but I don't know that the base rate is all that relevant. I'm not trying to claim that multiple authored papers are better than single-authored papers, just that an overwhelming majority of people's contributions (based on this very small sample) are based on multi-authored work–so people should resist the temptation to think of single-authored papers as better or more pure.
By the way, I once invited two people to collaborate on a paper of mine for purely strategic reasons: I thought the paper would get a better reception if it were associated with more authors. But the collaborators also improved the paper a lot.
I post again to explain better my feeling.
I am a student, so I am used to review the literature that most all the time comes pretty new to me.. I mean, the historical perspective that maybe you have, I think con't be seen by a young student.
Thus, very roughly speaking my hyphotheses are the following
1) Young Professors in statitists have, in general, an higher proportion of co-authored papers. This is due the PhD Thesis, that usually comes from a mentorship by more than one professor. Furthermore in applied statistics usually it means to deal with "multidisciplinary projects" . I have in mind a recent paper on the Annals, where a fresh phd student publishes with other 8 authors, that are famous Professors, Engineers, Industry Researchers etc. It's even true, according to what I see, that the job market even tends to give better chances to people that prove to be able to interact with "research teams" indeed co-authors ==> more than one paper.
2) I am not aware, (but I repeat I am a student so almost surely I can't see the entire panorama) of fresh phds or even young professors in a tenure track field that are able to publish consistenly on statistical journals sush as RSS-B, JASA meth. or Annals of Statistics with no co-authors. (I think that in the 70s or the 80s the situation was different.)
3) In Probability as well as in pure Mathematics the situation is different, because different is the kind of competion they face. When you deal directly with (just) "theorems" co-authorship is more dangerous.
By concluding the meaning of the my question has a personal appeal to me. I'd like to understand which path (try) to follow.
thanks for your attention
daniel
There is a paper in PLOS Biology on how to specify author contributions in a paperh. I like the approach and I see that PLOS is using some of this in their papers.
Gregor
Gregor,
The PLOS Biology article is interesting–but it's easy for biologists to be glib about this since all their journals have such high impact factors. Top stat journals have impact factors on the order of 1.0 or 2.0, much lower than the top journals in biomedical research.