Should ecology papers have guarantors who take full responsibility for a paper?

What does it mean for someone to be corresponding author on a paper? Does it mean they are taking full responsibility for the project, or does it simply mean that they uploaded the files to Manuscript Central? The answer to this question is important because authorship carries with it not only credit for a paper, but responsibility for it as well. At present, there is variation in what ecologists think is conveyed by corresponding authorship (more on this below). In working on a manuscript related to last and corresponding authorship practices in ecology, I have come across the idea of having guarantors of a manuscript — that is, one or more authors of the paper who are willing and able to vouch for the integrity of the project as a whole. This idea has been suggested repeatedly over the years (Rennie et al. 1997, Cozzarelli 2004, Weltzin et al. 2006) but has not been widely adopted. My goal with this post is to explore the idea of manuscript guarantors for papers in ecology, since this is the main point I’m stuck on with this manuscript.

First, to recap what the poll found: There was substantial variation in views on current and best practices for corresponding authorship. While most (54%) respondents said that the corresponding author “uploaded the files, managed the revisions and wrote the response to reviewers, and took responsibility for the paper after publication”, the next most common response (19% of respondents) was that the current practice is that the corresponding author is the person who simply uploaded the files. Those are pretty divergent views on the amount of responsibility someone is assuming by being corresponding author!

Blue bars indicate responses regarding current authorship practices, while gray bars indicate responses regarding best practices

One interesting result of the analysis is that people who received their PhDs more than 10 years ago were somewhat less likely to view corresponding authorship as taking full responsibility for a paper.

Thus, based on poll responses, at present it is not clear whether corresponding authorship is indicating that someone is taking full responsibility for a project. Given the ambiguity of the corresponding authorship signal, one way to indicate that someone is taking full responsibility for a paper would be to heed the calls of Rennie et al. and subsequent authors to have guarantors of a manuscript. When proposing this, Rennie et al. (1997) say:

All contributors are fully responsible for the portions of the work they performed and have some obligation to hold one another to standards of integrity. At the same time, special contributors must be designated and disclosed as guarantors of the whole work. Guarantors are those people who have contributed substantially, but who also have made added efforts to ensure the integrity of the entire project. They organize, oversee, double-check, and must be prepared to be accountable for all parts of the completed manuscript, before and after publication. In this way the role of the guarantor is precisely defined and differs from that of “first author” or “corresponding author” or “senior author,” there being many examples of these showing themselves unable to vouch for the whole work.

(Note: Rennie proposes doing away with “authors” in favor of “contributors”, hence the wording in the excerpt above.) I fully agree that current practices leave it unclear who is vouching for the project as a whole, and that having a guarantor system in place would encourage authors to think more carefully about this before submitting a manuscript. That would be a good thing. At the same time, the pragmatist in me wonders if this will really be adopted. I also wonder, for some complicated, multigroup publications, whether it will be possible for one person to be able to fully vouch for every analysis. For this, though, it could be possible to have multuple guarantors — for example, X is the guarantor for the empirical work and analyses reported in the study, while Y is the guarantor for the model and simulations. (I’m not sure that Rennie and colleagues would view a split like that as okay, though.)

All of which brings me to the thing that motivated me to turn this into a blog post: Do you think it would be useful to have guarantors for a manuscript? I will add a quick poll here, but I would love to hear more detailed thoughts in the comments.

Right now, this point is the only thing keeping me from posting my manuscript as a preprint — I can’t decide how strongly to argue for guarantorship for papers in ecology. What problems would having a guarantor solve? Would it create new problems? Feedback would be helpful as I think this through!

References:

Cozzarelli, N. R. 2004. Responsible authorship of papers in PNAS. Proceedings of the National Academy of Sciences of the United States of America 101:10495.

Rennie, D., V. Yank, and L. Emanuel. 1997. When authorship fails – A proposal to make contributors accountable. Jama-Journal of the American Medical Association 278:579-585.

Weltzin, J. F., R. T. Belote, L. T. Williams, J. K. Keller, and E. C. Engel. 2006. Authorship in ecology: attribution, accountability, and responsibility. Frontiers in Ecology and the Environment 4:435-441.

17 thoughts on “Should ecology papers have guarantors who take full responsibility for a paper?

  1. I think one of the problems here lie in that the responsibility is divided in different ways. I think senior author is always responsible for the work which is done in their lab, that is the integrity of both the data and the analysis, though they might have done neither.

    Then again, I don’t think there can be such single guarantor for the whole paper as it is quite a slippery slope as the contemporary science is full of black boxes. We, though, have a subtle understanding of how different parts of research project might contribute to the integrity of the end results and we acknowledge those in the paper: we used this one machine and say who manufactured it, we might thank lab manager who has arguable some level of responsibility that the data production goes well, we thank research assistants who helped to collect the data, we cite for the data we got from other sources (whether it’s weather data, satellite images or research data from other labs) and then we list what authors have done for the paper and who has been involved in any different parts of the project. It could be more formal, obviously, but what all needs to be guaranteed? I’m not going to vouch for my PCR machine, though it seems to every now and then produce reliable results.

    Then again, what “accountability” means here? Legal accountability is quite well defined and it’s useful when things go very wrong but otherwise, what kind of accountability is needed? My view of the corresponding author’s role is that they are the one who knows where to forward the questions about the paper.

    • All good questions and points! One thing I’ve wondered about is, if I was clicking a box saying I was a guarantor for the manuscript, would I do anything differently? For example, as you say, the senior author already is viewed as being responsible for the data and analysis collected in their lab, even if they didn’t collect the data or analyze it themselves. I tend not to review the code my students use to analyze their data — I ask lots of questions if a statistical result doesn’t seem to match what the figure shows or if the number of data points on a figure doesn’t make sense or something like that, but I don’t have a formal code review policy. Would I create one if I was clicking a guarantor box? I’m not sure, but it’s something I’ve wondered about as I try to think through the guarantor idea.

  2. I wasn’t and still am not quite clear on what you mean by “responsibility for the paper post-publication”. That could mean simply being the one who takes responsibility for sending out PDFs, fielding questions, and promoting on social media. Or it could mean being the one who takes full blame if it turns out later some other author faked something (or made a mistake). Those are really different things; when I answered the poll I was leaning to the former, but this post now makes me think you meant the latter. If so – that seems like a role for the senior author, if anybody. “Corresponding” to me is simply an administrative role.

  3. I’ve done some statistical consulting in the world of medical communications, and in the health sciences it’s incredibly common to have guarantors of the data, analyses, etc. I imagine that this is for legal reasons? (Me making an honest mistake on a macroecology paper -> lots of embarrassment for me, a little bit of embarrassment for my supervisor, probably no harm done. Someone like me making a mistake on a paper for a pharmaceutical drug company -> serious real-world implications.)

    I’m not opposed to introducing this tradition in ecology, but we’d need to have precise definitions. There’s always going to be some sort of error rate — a number gets typed in wrong, a command in R doesn’t do what we think it does, a fancy machine gets miscalibrated. Ideally we catch most of these before the paper’s published, or at least conduct our analyses in an open way so that others can see what we did, but it’s hard to take full responsibility for something someone else did, particularly if we’re borrowing terminology that has legal ramifications in other fields.

  4. I voted “no”, but only because I already view the corresponding author, who is often also the first author, as the guarantor. It is now obvious (thanks to your poll) that people view corresponding author in many ways, so maybe I should rethink my position. In the end, I think that adding the “guarantor” designation would just be another tick box that doesn’t change anything, hopefully because first or corresponding author is already doing the right things. I know for my first-author papers I would just check a guarantor box along with corresponding author box without thinking twice.

    • Yes, the pragmatic side of me wonders how much this would really change things. But the wide variation in views on corresponding authorship makes me think some better way of indicating responsibility might be worth trying.

  5. Fascinating question to ponder. Having read a decent bit recently on how people’s minds work, my prediction would be that having to check that box would almost certainly change people’s behaviour. I agree with the comments implying that it’s often essentially impossible for one person to vouch for every detail (imagine a study with some molecular genetics, soil chemistry, and Bayesian modeling). But signing your name as guarantor seems likely to prompt an increased effort to scrutinize the parts you’re not expert on – usually via more thorough discussions with the technical experts, I imagine. It’s not that project leaders don’t already do this, but it’s not black-and-white – I think it would push forward the line of what constitutes “enough” double checking, at least on average.

    One side benefit might be to get everyone to slow down a bit, shifting the incentives back, at least bit, from quantity towards quality.

    • I think the biggest benefit would probably be that it makes explicit that authorship is not just about credit, but also responsibility. I agree that it would probably influence behavior. Of course, I suppose there’s the chance that it would make the non-guarantor authors think they don’t bear (as much) responsibility as they currently do, which would definitely not be a good outcome.

  6. I agree with what other folks said. I don’t understand what problem this “guarantor” thing is supposed to solve. I don’t see how it would alter anyone’s legal responsibilities or affect the outcome of any misconduct investigation. (And if that’s what it’s supposed to do, a *lot* more thought needs to be put into definitions of duties, enforcement mechanisms, penalties, etc.) And I don’t think it would change the behavior of a sufficient number of authors sufficiently substantially to reduce rates of either honest mistakes or misconduct. In part because those rates are already low.

    As for whether just having the “guarantor” checkbox would alter author behavior in some more nebulous but still-useful ways–making authors more conscious of their responsibilities, say–I question whether it would. For instance, like Meg I’m not at all sure I’d do anything differently than I already do if I had to check that box. And if you say that that means I’m already doing my job as “guarantor” and that the checkbox isn’t aimed at people like me, it’s aimed at less-responsible PIs, sorry, but I have yet to see a case in which obliging less-responsible people to check a box on a form has made them appreciably more responsible. In my experience, people who become conscious of their responsibilities in response to seeing a checkbox on a form are invariably people *who were already acting responsibly anyway*.

    We have other evidence for this besides my own anecdotal experience. For instance, many journals have policies on who can be an author, policies that forbid some increasingly-common authorship practices. According to our poll data, those policies are mostly ignored by authors.

    But I’m open to being convinced otherwise. My admittedly-sketchy understanding of the behavioral economics literature on “behavioral nudges” is that they have very small effects on individuals, but can have substantial collective effects if they’re applied to *huge* numbers of people. So I dunno, maybe whatever tiny effect a “guarantor” checkbox would have on individual behavior would scale up to a non-negligible effect at the level of science as a whole? (https://dynamicecology.wordpress.com/2015/11/05/whats-a-small-effect-anyway-and-when-are-they-worth-caring-about/)

  7. All authors should be confident of the results, but when they come from very different areas, no author can be positive there are no problems. If my paper has some GC/ms in it, I cannot verify that it is correct except to the extent that I have confident in the collaborators. All these concerns are just about people desperate for credit. The more authoritarian labs have the old, usually male, PI take the senior, last position, and the corresponding position, though he usually would not even be able to upload the manuscript, or vouch for any of the techniques. These things are all conventions and these days interested people can correspond with whichever author they feel like. Emails are easily found. So the argument that the corresponding author should be the one in the most stable position no longer holds. Still, in biochem and cell bio it is more the convention that the last author is the lab head and the corresponding author. In ecology and evolution, it is more convention that the first author is corresponding and the last is the lab head, provided she actually contributed to the paper. No one can vouch for everything on the best, most collaborative papers.

  8. This obviously intersects a lot with Greg’s argument last week about responsibility being a component of authorship, with which I agree. So in the language of this post, to my mind anybody who is an author is a guarantor (and I don’t think its realistic to have just one guarantor).

    As an EiC I deal with cases of erroneous papers and plagiarism and salami publishing (I haven’t dealt with outright fraud yet) and you better believe it reflects badly on every author on the paper (all of whose names will go on the retraction). Yest it reflects most on the first author and senior author, but those 5 names in the middle are swept up too.

    So please don’t put your name on a paper if you’re not confident it is a good faith effort and willing to stand by the paper if it is questioned. Or to expand on the Oikos criteria that was mentioned in Greg’s post – you shouldn’t put your name on the paper only if you’re wiling to present and defend it in public but if you’re willing to defend it if the EiC gets a formal challenge to its validity. Such challenges do happen and should be on people’s minds when they accept co-authorship.

    This is the dark side of the ever increasing number of authors per paper (people put their names on papers they have no idea about the quality of), and it is a good reason to put a cap or even deflate that practice a little.

  9. Here is an answer from a friend of mine, Liz Haswell, who is a plant cell biologist. She points out that the lab head would also have reagents and that is an important part of cell biology papers and communication. She gave me permission to post her answer in full:

    Hi Joan,
    In all my papers from Wash U [her first faculty position] I am last and corresponding!
    Yes, I do all the uploading but to me the corresponding author is the one that someone else would contact with questions or reagents. I’m the stable person and the financial sponsor so I’m last and corresponding. If a postdoc was leaving with a project, plants, and plasmids, and knew more about the project than me, I’d have him/her be corresponding but that hasn’t happened yet!
    Liz

  10. Responsibility does not mean that the leading (or corresponding, first, or last, depending on the field….) authors can check or replicate everything in the paper from the scratch (without their coauthors) but simply that if some responses or critiques arise, they coordinate, respond, and will do best in addressing the concerns – ultimately to clarify or actually admit if errors are found (or to provide codes or data etc.). Of course, under normal circumstances, coauthors who did specific parts of the work will help or take over with such issues. But leading authors cannot behave asymmetrically – if everything goes well, they take upside from being the leading author, but if some errors are found, they cannot just shift the blame to the author who actually mis-code an analysis and somehow think that they are not responsible. So they must be ready to take downsides also (e.g., simply admitting that errors happened) – although we all hope we will not face such situation.

  11. Interesting question. I think guarantorship seems redundant because the code of authorship implies that you are already guarantor if you’re an author. I think adding a double-bagging layer just creates another level of unnecessary administration & also could potentially raise legal issues. Is guarantorship more legally-binding than co-authorship if one of your co-authors who was responsible for one part of the study is found to have done something dodgy?
    Also, a couple of people here have said that corresponding author is just an administrative role. I was under the impression it is much more than that – sure, most people only use that email address to get a pdf, but isn’t it meant to be a point of contact for anyone (e.g. media, potential collaborators) who wants to ask further questions about the study or potentially develop a new collaboration from the data? How many people use the corresponding author email in this way? As corr. author, I’ve only ever been contacted by others for pdfs. But I have personally used the corr. address to contact other authors if I was interested in clarifying aspects of their study that weren’t explained in detail. I see the corr. author as the person with the long-term responsibility for the paper.

  12. This is my first comment here on the blog, as I was unable to express my opinion in a single tweet (sorry for any confusion arising from that). I should first say that I really like this blog, especially the current series on publishing/authorship, and the insightful polls. In this particular case I would not agree though.

    As I understand the reasoning here, one person could take all the blame, regardless of whether it is her/his fault (yes, I deliberatle put females first, because I expect a gender bias). Such practice is actually already in place in the author guidelines of Ecology Letters: “Some authors, especially the lead author, are responsible for the integrity of the entire manuscript”. But I do not think this fair. Shouldn’t the last author be blamed for lack of oversight, if e.g. the study design is completely wrong (e.g. pseudoreplicated to N=1) or adresses a question that has been solved years ago? If we make the last author responsible for everything, on the other hand, we may end up protecting someone who provided fraudulent data.

    I can think of only two ways of making this system work: First, the responsibility is split among multiple guarantors (as mentioned in the text), so for every part of the paper there is one person nominated who takes the blame. This solution sounds to me like the “authorship contributions” section. On a recent (well, relativley recent) poll ( https://dynamicecology.wordpress.com/2016/08/08/views-on-authorship-and-author-contribution-statements-poll-results-part-2/ ) there is no inclusion of the statement “I read authorship statements to detect who is responsible for mistakes” , so authorship statements are probably not needed to sort out problems of responsibility. Anyway, I do not see a benefit in renaming it to “guarantors section”.
    A second solution would be that the guarantor takes the position of a mediator in case problems arise: He (or they) should make sure that the right author is blamed if problems arise. But this regulation is already in place in the ICMJE guidelines, which list as author requirement:

    “Agreement to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately *investigated and resolved*.”
    Thus, it is the already the responsibility of all authors to name the person who is to blame for it.

    Overall, I think that the guarantor system is redunant, as all the functions are already provided by the authorship section and the corresponding authorship.

  13. I voted “no” as well, but perhaps for reasons others have not expressed. When I think of the term “guarantor,” I naturally think of “guarantee”. To me, a guarantee in a commercial/ public sense is something issued by either the entity manufacturing the product, or the entity making it available for public consumption, or both. So for instance, I purchased a flying drone recently for my son’s birthday. The product worked as advertised- for about 3 weeks, and then developed a grinding noise in one of its motors.

    We returned to the retailer, who initially said all product defects, and related returns/ replacements, were handled by a manufacturer in China… You can probably envision my eyeballs rolling at that point of the conversation. My boy was also glum about the ordeal- realizing it might take months to replace his flying drone. The clerk, sensing our disappointment, then said her store guaranteed the product on the basis of satisfaction- i.e., if we were not satisfied with its performance, “as advertised in the store,” then the store would guarantee the product and replace or refund our money. We happily left with a new drone that has functioned well since.

    And so I voted “no” in your poll not because the idea of a guarantor is a bad one, but because it seemed not quite complete. I think that if we require one or more authors to register as a guarantor, then the journal’s editor(s) should do the same. While it is not possible for an editor to catch every last flaw or potential fraud in a manuscript, clearly some journals publish more drivel than others- and that is a result of poor editing and/ or peer review. I like the idea of holding the authors’ feet to the flames, as it were, but also holding to account the journals publishing the material.

    Indirectly, I also believe that some of the issues highlighted by your post might lend some credence to the notion of open/ public-disclosure peer review. While many issues have yet to be resolved for this kind of format, I think in the long run it solves many of the problems we observe in the publication process. Were authors to make available their manuscripts and related data for public scrutiny- prior to publication, we likely avoid many of the problems that might necessitate a guarantor in the first place.

Leave a Comment

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s