Thursday, April 08, 2010

Playing the Game (2)

Citation index considerations aside, have you made any publication decisions specifically because you wanted to increase your number of publications, on the cynical-but-perhaps-justified assumption that more papers = better?

For example:

1. Have you taken what probably should have been one paper and split it up into two or more shorter (but nevertheless distinct) papers for the sole purpose of increasing the total number of papers on your CV?

Figuring out how many papers should be written about the results of a particular research effort is a bit of an art, and in some cases more papers is the best decision. Sometimes you know in advance whether to write results up as one or more paper(s), and sometimes you don't know until you start writing.

My question, therefore, refers to cases in which the main motivation for splitting up a project into n>1 papers is to crank up the publication numbers.

2. Have you published what probably should have been one paper but that instead were two or more related papers that were not as distinctly separated in content as in the first scenario? (i.e., shingling)

This is also a fuzzy concept because sometimes you want to publish one short and zippy paper and then another, longer one in a more specialized journal. Is that shingling? Or what if you decide to publish some different-but-related papers because each involves different groups of co-authors and it makes logistical sense to keep the publications separate? Or maybe you want to publish a review paper that summarizes information in some of your other papers. There are many legitimate ways in which similar papers are published by the same author.

But then there are other cases. I recall one time when I brought a manuscript to review on an airplane. This was back in the Days of Paper, so I had a hard copy of the manuscript, and had turned to a page with a figure on it. I was traveling with some colleagues to a conference, and a colleague in a different-but-related field was sitting next to me. He was also reviewing a manuscript, and turned the document to a page with a figure on it. It was the exact same figure. Yes, I know that manuscripts in review are confidential, but there we were, sitting next to each other with two different manuscripts submitted to two different journals at the same time, and there was at least one thing in both manuscripts that was identical. So we started comparing. The two manuscripts had the same authors, the same figures using the same data for the same topic, and only a slightly different "spin" put on different aspects of the data. Those two manuscripts clearly represented shingling of a sort that was probably not OK. We informed the editors of the journals.

3. Have you ever submitted a paper before the project had advanced as far as it probably should have before writing up part of it as a manuscript; i.e., a premature manuscript that was probably publishable but that you knew would be much better if you worked on the research more? (but you didn't feel you could afford to wait longer because you needed publications on your CV sooner rather than later)

As with the other cases, this is not an obviously bad thing to do either. Maybe you would have waited longer to publish if you already had tenure, but there are also good aspects of publishing a preliminary paper to communicate initial results rather than waiting, perhaps years, to publish one big definitive paper. You'd want to be as confident as possible that the preliminary paper was sound, but if you feel you have something to say that is of interest, I think it can be very useful to publish early and often.

I know that that is "playing the game" and perhaps contributing to the mass proliferation of academic articles so that the flow of information is overwhelming and the very act of publishing a scholarly work is devalued etc. etc., but I think that if you have something interesting to say, it's a good thing if you write it up and get it out there.

Ideally, in the course of your career, you will publish some short papers and some longer, more detailed papers and some review papers and some papers in Awesome Journal X and some other papers in Specialized Journals Y and Z, and it will all even out. It's in the early stage of an academic career when every decision about what/where/how much to publish can seem so critical.

From what I've seen in the physical sciences, the "best" route to take at all stages of an academic career is to try for a balance between publishing a reasonable number (according to the norms of your specific field) of very good but perhaps not awesome papers in respected journals, and then some (but likely fewer) rather awesome paper(s). This is preferable to having lots of narrowly focused papers or only/mostly having "big idea" papers. Together, however, these different types of publications demonstrate the breadth and depth of your research.

When deciding how to divide up a big project into papers, I try to optimize publication quality, speed, and impact, as well as consider what is most fair to the most number of co-authors, particularly students, postdocs, and/or tenure-track colleagues. This not always a straightforward decision, of course, and sometimes I wonder if I would do just as well by consulting a Magic 8-Ball for advice.

Or maybe I should go back for yet more training in Responsible Research Conduct. Surely somewhere in all those case studies and PowerPoint presentations, there is a nifty formula for sorting all of this out easily. {note use of delusional font}

19 comments:

  1. I'm a fourth year graduate student working on a project that might well be best published with significantly more data, data that would have to be collected in the time after I graduate. But this is my baby - I've been working on it from the start, and in this case, I favor my own career prospects over what might be the absolute best way to record and put into context my work, and I don't think many people would begrudge me that. In other words, I think it's expected that human factors come into the publication process, and that's probably OK.

    ReplyDelete
  2. 1 - No
    2 - No
    3 - Definitely, but in Conferences.

    In the field of Computer Science, publishing in Conferences is more considered that in other fields, such as Mathematics or Physics for instance. Most Conferences are peer-reviewed and the top-tier are harder to enter than any journal of the field, because they select as much and the authors do not have an occasion to answer the referees.

    My understanding has always been that conferences were adequate for "talking about your results" and that you could present a result before its full maturation. A journal publication, in return, comes several years later, is more mature and sometime gathers the results of several "conference papers".

    In practice of course some conference aims to be as good as journal, and the "publish or perish" policy pushes (computer) scientists to publish one journal paper for each 10 page conference paper, basically giving the same results but with more details given the extended space allowed by journals.

    I am curious of how it is in other fields?

    ReplyDelete
  3. Some of my thesis work had better luck being accepted in computer science conferences after we split it up into multiple papers. Perhaps there were too many distinct ideas for the reviewers to take the single paper. So that's sort of a case of (1).

    Also, conferences have deadlines, like jyby wrote, so that one is often submitting to conferences before one would like.

    ReplyDelete
  4. I published a paper in 1970 which many of my colleagues considered to be of fundamental importance. I published it in the proceedings of the conference where I presented the results, a completely unrefereed publication. No one had explained to me the importance of refereed vs. non-refereed publications while I was a grad student; I was an associate professor by the time I understood it. I have been urged over the years to republish this work in a refereed journal, but I decided that publishing the same thing twice was not really ethical, so I have never done so. It is still available only from me personally, or from the somewhat hard to get Proceedings of the Emory University Topology Conference of 1970. And, it is still regarded as very important by some colleagues.

    ReplyDelete
  5. I have done the opposite, published one paper when it should have been two, a biology oriented one, and one about methods development.

    I was frankly too lazy to write two papers (it was in the last months of my postdoc), and I still regret that the methods development part doesn't get the exposure that it deserves, in my opinion.

    ReplyDelete
  6. 1. No
    2. No
    3. No

    I am a fifth year grad student in astronomy. In hindsight, I wish I had divided up a big paper I published that had a clear way to separate into two parts. I think it would have increased my job prospects since number of papers is the primary metric used to evaluate postdoc applications. So, I wish I did say yes to #1, and I will probably keep this in mind in my future work.

    ReplyDelete
  7. "Have you ever submitted a paper before the project had advanced as far as it probably should have before writing up part of it as a manuscript; i.e., a premature manuscript that was probably publishable but that you knew would be much better if you worked on the research more?"

    Everyone in biomedical science has become terminally frustrated by the review process. in virtually every journal, regardless of where they stand within the "hierarchy", because essentially every reviewer sees it as his or her responsibility to tell the authors what to do next. This is regardless of whether the paper has little or lots of data. There is no "standard" for a complete story--the standard is "Whatever they submit plus one more year's worth of work that I can suggest "

    This is leading me, always someone who loved "complete stories", to strongly consider submitting papers that are "incomplete", and to even think about leaving out data that is already in hand to be able to add it after revision.

    On the more constructive side, i am lobbying colleagues to THINK before they ask for more experiments--does this meet the standards of most papers published in this journal already?

    Mark P

    ReplyDelete
  8. From an editor: Referees ALWAYS suggest new experiments. We think pretty carefully about whether the new experiments make sense (i.e., provide critical support for the central hypothesis) or are just random extensions or somewhere in between. We probably end up annoying both parties (the authors have to do more work, yet it's often not as extensive as what the referees wanted), but at least it's on our minds!

    ReplyDelete
  9. I am about 5-8 years behind in writing up my research results---I definitely prefer more complete stories over the least publishable units.

    I'm one of the reviewers for Faculty of 1000, and I've found it very difficult to find papers to recommend. Almost all the papers I've read lately have had good titles and abstracts, but huge holes in the science, with sweeping conclusions based on bad experiments (or, sometimes, nothing that even resembles an experiment).

    So, unlike Mark P, who said
    "On the more constructive side, i am lobbying colleagues to THINK before they ask for more experiments--does this meet the standards of most papers published in this journal already?"
    I'd like to urge authors and referees to THINK---does this work meet the smell test? or is it really shit?

    ReplyDelete
  10. I have occasionally split a paper that would have been better as one, but only to resolve intractable authorship problems. In one case including one additional figure (which would have been nice to have) would have meant picking up 15 additional co-authors, so the whole thing was split into two papers in order not to dilute the work of the lead author grad student.

    ReplyDelete
  11. EuropeanFemaleScienceProfessor4/08/2010 04:51:00 PM

    Don't remember if I commented on this here or not. In Germany, the DFG now insists that only 5 top publications be mentioned. Full stop. For a report on funding 2 publications per year of funding are the max. So this means you have to carefully select what you put in.

    I'm suggesting that our search committees also use this method and save ourselves from endless bragging and fake publications.

    ReplyDelete
  12. A company I worked for had a string of patents developed with a non-tenured professor. The patents could have been combined into one, but the venture capitalists insisted on splitting in to about 20. Instead of just patenting the general algorithm we patented each possible variation separately. The professor did not enjoy explaining that to his tenure committee.

    ReplyDelete
  13. A year and a half ago I published some results in a 3 page Letter in a good journal. I knew the project was only part finished, that there was more to be found. But I also knew that the field was moving fast, and that what I had worked out was still an important piece of the puzzle. Since then I've discovered more, and I've realized that I could have written a longer and more complete Article for an even better journal.

    OTOH, I don't regret not waiting, and not just because it got me a paper at an important point in my career. Getting this result out there brought me to the attention of some other people, and through interactions with them I've learned useful things that have helped me with the work.

    So I say go ahead and publish the LPU now and then. If done right, it can actually make for better science, as it puts you on the radar of others and lead to useful collaborations.

    ReplyDelete
  14. Nope I've never done any of those things you wrote about in your post even though I was extremely aware that this is what other people do all the time. I don't think highly of people doing this, so I refrained from doing these things myself with my own papers. And that is probably why I didn't get very far in my academic career as my output (in terms of pure quantity of papers) doesn't match that of my peers.

    ReplyDelete
  15. publishing rapidly and prematurely, or publishing LPUs, in order to move quickly to avoid getting scooped, is something that is common in my field. However my opinion is that if the work isn't completed, there's really nothing to be scooped on in the first place. having an idea or a suggestion of data is not the same thing as having concrete and rigorous data

    ReplyDelete
  16. I think that your questions cannot just have a yes/no response because for the vast majority of the papers (for eg. in cell biology), findings in each paper throw up many more questions. One has to draw a line somewhere as to what is 'complete' and leave out some questions for the future. Thus, the yes/no response becomes dependent on each individual's idea of completeness of a study.

    ReplyDelete
  17. another European FSP4/09/2010 06:06:00 AM

    1. Yes. In my field it's much easier to publish shorter papers, so if a paper naturally splits in two parts I will usually do so.
    2. no
    3. no, but I have given talks about unfinished stuff, clearly declaring its unfinished (=may never be finished) status. That's what in my field everyone does if the conference comes and your result isn't ready.

    @EuropeanFSP: I wish every agency insisted on having only the best publications evaluated. There would be so much less trash (and my cv would look SO much better).

    ReplyDelete
  18. "I know that that is "playing the game" and perhaps contributing to the mass proliferation of academic articles so that the flow of information is overwhelming and the very act of publishing a scholarly work is devalued etc. etc., but I think that if you have something interesting to say, it's a good thing if you write it up and get it out there."

    Very big point! I definitely feel information overload and have issues keeping up with the literature. I always worry that I'm missing something out there. Maybe you could write an article about what technology tools you use to keep up with the literature? And how you manage all the information? I think this is an important topic that is often ignored in the advisement of graduate students, to our detriment, now that there are so many journals.

    ReplyDelete
  19. I agree with both Mark P and Kevin.

    Another year of experiments won't make a pile of shit into something better. It just makes a bigger pile of shit.

    Conversely, delaying a good story by a year or more might just be the end of someone's career.

    ReplyDelete