« Animal Spirits | Main | Columbia Business School has a blog »
Why are there so many co-authored papers?
A loyal MR reader asks:
I’m reading (grr) a lot of academic papers lately and, to keep myself awake, have wondered: why do they almost always have more than one author? Is it like cops in New York City—they’ve got everyone persuaded it’s too dangerous to go alone? Is there some networking benefit, professional or psychological? Does it just enable everyone to claim more publications? Has anyone studied which fields have the highest and lowest average number of authors per paper?
I thought you could blog something interesting on this. I might add that the papers couldn’t be any duller, and I wonder if committee authorship plays a role in this as well.
I believe that co-authored papers are correlated with:
1. The existence of a laboratory
2. Senior scholars who generate funding and thus gains from trade
3. Empirical work, which tends to be more divisible than theory; co-authored papers are relatively rare in pure economic theory and in philosophy
Co-authored papers are becoming increasingly common in economics, also because the effort requirements for top publications have been rising. In most cases a co-authored piece is worth at least 2/3 of a singly-authored piece, so the incentives for co-authorship are strong. Here is an earlier post on co-authorship.
Posted by Tyler Cowen on January 29, 2008 at 06:54 AM in Science | Permalink
Comments
#2 reminds me of popular novelist James Patterson, who has been releasing a stream of co-authored books lately. Some cynics wonder if he has written much more than his name on the cover and the check endorsements.
Posted by: burger flipper at Jan 29, 2008 7:09:02 AM
Certainly labs make a difference, but much of it's related to funding. It's almost impossible to get grants as an individual in any area classed as science (and that's a broad definition) and many funding bodies also want multiple universities (and often countries) involved. Hence the research is being carried out by a team of people. Publications tend to follow on from this, with team members allocated to various papers, often a lead researcher with numerous post-docs, although there are other models. Another reason is that for many of us it's more motivating to work with others. It makes it into more social rather than solitary work, which it's easy to put off when writing alone. But with other authors there's more pressure! And it's good to share the workload anyway. Not sure if it correlates with dullness - in my experience joint academic authorship does not function like committee authorship! If anything, I'd say the opposite, with colleagues encouraging bolder statements.
Posted by: Dave Rowley at Jan 29, 2008 7:38:40 AM
A. Smith suggests: "The division of labour, however, so far as it can be introduced, occasions, in every art, a proportionable increase of the productive powers of labour."
Posted by: Mike/KP at Jan 29, 2008 7:47:56 AM
While studying open source and patenting, I took a look at the hyperauthorship phenomenon in high-energy physics and genomics. In those fields, it is not uncommon to have papers with 50+ authors. It is related to the expense and manpower required to operate such massive facilities.
Posted by: Colin at Jan 29, 2008 7:54:09 AM
In my experience, single authorship is the norm in the humanities. Sociology has less co-authored papers than economics and psychology, which in turn have less than the natural sciences, where single authorship is rare.
All of which fits nicely with your hypotheses above.
Posted by: LemmusLemmus at Jan 29, 2008 7:55:49 AM
One reason is working outside your field.
I have many jointly authored papers, mostly resulting from the following scenario. Someone at a conference proposes a problem. I solve it quickly, but I don't know it's significance. I also don't want to steal it from them.
Co-authorship is a perfect solution. They write the intro and explain why it matters. They usually also suggest improvements/extensions, and whack an easy mole or two (moles that I've never heard of).
Not to mention, double the publicity.
Posted by: mathgeek at Jan 29, 2008 7:57:44 AM
"In most cases a co-authored piece is worth at least 2/3 of a singly-authored piece, so the incentives for co-authorship are strong."
Is there some objective valuation behind this statement?
Posted by: Billare at Jan 29, 2008 8:04:04 AM
If anecdotal evidence can be counted, the number of co-authored philosophy papers is rapidly rising. This is true even in those sub-fields that are not employing a lot of scientific concepts.
Math has gone from a single-author field to a multi-author field over the past generation or two.
Posted by: Justin at Jan 29, 2008 8:06:00 AM
Don't forget the grad student - advisor dynamic.
Posted by: JB at Jan 29, 2008 8:33:49 AM
Most papers in science and engineering list the supervisor-professor and the graduate student or students that did the majority of the work and writing. It would be considered academically dishonest not to list the graduate research assistants who worked on producing the paper, though I understand that at least in legal academia this isn't the case (having had a friend or two do such work for a lawprof and not get or expect co-authorship). Even purely theoretical papers in math, physics, and engineering are usually co-authored with the graduate students.
Posted by: billb at Jan 29, 2008 9:26:01 AM
from my experience, it is both more fun and a better way of learning to discuss your ideas with a colleague (who may give you a different perspective) and look for solutions together. Hence, I prefer to co-author my papers.
Posted by: lb at Jan 29, 2008 9:38:47 AM
Also, since I just noticed the penultimate sentence, I thought I'd point out that there's no loss in value in science and engineering due to co-authorship. In fact, I tend to look slightly askance at singly-authored papers since it could imply that the author has no students or that the author failed to credit the grad students that worked with him. The former isn't a disaster, but it could mean that the author isn't contributing to grad student training. The latter is a cause for action.
Now, there is some bias toward granting more worth to papers that an author is the lead author on rather than a second or third. This is less important for professors late in their careers who have many grad students to supervise than it is to professors early in their careers who could be expected to have fewer grad students and to be doing more of the work themselves.
Posted by: billb at Jan 29, 2008 9:52:08 AM
Laboratory and empirical work both imply you looking for a "right" answer, so if you make a mistake your result will likely be proved wrong at some later date. Two or more people working together are less likely to make embarrassing mistakes.
Posted by: joan at Jan 29, 2008 10:17:30 AM
Some of the comments here from researchers with a science background might not be aware that in economics, at least, the concept of a "lead author" isn't really known. In fact, the order of names is often completely random, and, as far as I know, the credit given by tenure committees and the like is equal for all authors. This may be part of the reason that economists typically only credit those who made substantial contributions to the theory of the paper, and note only in footnotes assistants and grad students who wrote code or solved a proof here and there.
Posted by: cure at Jan 29, 2008 10:37:29 AM
Outside the halls of academia it may be seen by some companies as a way to increase the worth of their employees in the eyes of the client and to provide an instant patina of expertise.
Posted by: Pitt at Jan 29, 2008 11:01:01 AM
Co-authorship often produces papers that could never be written by either author alone. Mathematics has a fine history of this, with two or more people having complementary skills collaborating to work out the details of a theory. Just look at the work of Paul Erdos, who made an artform of mathematical collaboration.
Posted by: Dolohov at Jan 29, 2008 11:04:48 AM
Multiple authorship is not a new phenomenon, although the increases in the number of co-authors is somewhat amazing (and apparently also a possible source of research fraud, but that's another story). It does have a lot to do with the sociology of the particular discipline. Isaac Asimov tells the story in the first volume of his autobiography of working really hard one summer to complete the research for, and write, a paper for a chemistry journal, which was, in fact, published. And for which he caught a lot of flak from his colleagues--single-authoring was just not the way things were done. In, as I recall, 1948 or so. In chemistry.
Posted by: Donald A. Coffin at Jan 29, 2008 12:05:01 PM
There's also the problems of co-authorship in papers that are mere shadows of papers published in other journals. So if four authors co-write a paper that get's parsed out into four articles with some or all, or even different co-authors, just how much research is really there?
From NATURE: "As many as 200,000 of the 17 million articles in the Medline database might be duplicates, either plagiarized or republished by the same author in different journals...
Mounir Errami and Harold ‘Skip’ Garner at the The University of Texas Southwestern Medical Center in Dallas, used text-matching software to look for duplicate or highly-similar abstracts in more than 62,000 randomly selected Medline abstracts published since 1995."
And that's just Medline, not social science or econ databases.
Posted by: The other Eric at Jan 29, 2008 12:24:08 PM
By "co-authored," I presume you mean 2 P.I.s?
As in, not "co-authored" as in the graduate student who did all the work and his advisor, right?
Posted by: Andrew at Jan 29, 2008 12:44:14 PM
You need to be known to be respected...Ergo, you need social standing. How best to get that social standing? Piggy back on someone else's and borrow it until you get your own. Take a look at the credentials for many co-authored papers in humanities and the social sciences and you'll find a bunch of graduate students piggy backing on the reputation of the one or two tenured faculty members also listed as co-author. Call it mentoring or nepotism, take your pick, but it works.
Also too there is the old publish or perish at work. Overworked up-for-tenure faculty member with more students then time, who nonetheless needs to get a few publications listed to his name in the next few quarters. Easiest way to do that? Lend their name to some students, who do the actual legwork, including writing the paper. The faculty member edits it and "whacks a few moles" as someone else here put it, and then uses personal influence to get the thing published. Far easier on the faculty member.
Extremely rationalized division of labour indeed!
Posted by: Tamara at Jan 29, 2008 12:46:05 PM
"And for which he caught a lot of flak from his colleagues--single-authoring was just not the way things were done"
Single authorship might imply being hard to work with. But a single instance of it is hardly a trend. Cooperation can be good, not always, and should be used on its merits, not encouraged beyond its benefits. Solitude can be good, too. The right tool for the job should be found over time through experience, not heavy-handed arbitrary incentivization.
It annoys me when people make virtues of necessity and vice versa.
Posted by: Andrew at Jan 29, 2008 12:53:34 PM
An expert and his programmer. Often programmers develop useful intuitions by doing the problem and wind up an equal author in return.
Posted by: Ron Hardin at Jan 29, 2008 1:10:20 PM
Might "fun" and "ease" have anything to do with it? At least in some cases?
FWIW, I've co-authored some long-form fiction, and I loved the process: No loneliness, both of you jogging each other out of delusions and mopes ... Loads of fun.
In fact given what a drag it can be to sit alone for months in front of a keyboard, I'm amazed that more people don't co-author long-form fiction. That does depend on personality and chemistry, of course.
Posted by: Michael Blowhard at Jan 29, 2008 1:21:08 PM
In Cognitive Psychology and Neuroscience, multiple author papers are the norm - with multiple experiments, often designed and run by different collaborators are combined into larger papers.
Posted by: Paul at Jan 29, 2008 1:22:48 PM
The simplest explanation: papers with multiple authors are higher quality and so are accepted more often. This isn't too hard to believe, since each author can focus on different aspects of the work, and more eyes means more bugs and errors are found. I agree with previous comments that there is a division of labor advantage going on.
In computer science I would say the norm is 1-4 authors per paper. The best papers tend to have 2-3 authors. Single author papers can be good, but often lack polish or completeness.
Posted by: Nathan Whitehead at Jan 29, 2008 1:35:10 PM
