Friday links: a striking correction in ecology, the greatest female scientific illustrators, and more

Also this week: individual responsibility vs. climate change, and more.

From Jeremy:

A correction for Start and Gilbert 2017 EcoLetts. I link to this because some of the issues it corrects aren’t the sort of thing one usually sees corrected. In particular, it’s not every day you see a correction that says (in so many words) ‘Organisms were not assigned randomly to experimental units, as stated in the original paper. Actually, they were assigned to experimental units in a maximally non-random fashion.’ The strange thing is, the correction states a scientific rationale for the non-random assignment. So why wasn’t it stated in the first place? The correction addresses many though not all of the issues raised in the PubPeer comment thread about the paper.

An interesting podcast conversation (and transcript) with historian of science Patricia Fara. Covers the greatest female scientific illustrators, Isaac Newton at the Royal Mint, how the Scientific Revolution wasn’t actually a revolution, what we know about ancient science, why basic geology took so long to get right, the history of taxonomy, and more.

Jessica Hullman on the challenges of choosing one’s scientific methods and practices so as to send the right “signals” to others. I’m not quite sure what to make of this, and worry that I’ve misunderstood it. Personally, I’d recommend just doing science as best you can, and not worrying about any implicit signals you think your methodological choices might be sending to anyone else. But YMMV, apparently.

Philosopher Säde Hormio on individual responsibility for the “structural” harms of climate change. I don’t completely agree with this. I think it’s too confident our ability to infer the small and indirect effects of individual choices, and insufficiently concerned with the effects of our choices on other things besides the “structural” harms of climate change. But I wanted to link to it, because it had me thinking back to this very good old post by Mark Vellend, and the associated comment thread.

9 thoughts on “Friday links: a striking correction in ecology, the greatest female scientific illustrators, and more

  1. The unwillingness of editors/publishers to retract is ever amazing. Maybe Start and Gilbert’s study was solid, but there should be some penalty for saying treatments were randomized when they were targeted, presumably to improve odds of getting effects. And then dang the luck! Someone stole the first authors computer which had the only copy of the data so he can’t post it. There were two authors but only one copy of the data, no OneDrive, GoogleDrive, thumb drive or anything else. Pretty easy for Ecology Letters to make a non-judgmental retraction. 1) Methods were fundamentally different than described, and 2) missing data, against policy. Retract and go publish it elsewhere. But no, a correction is fine.

    • A few possibly-pendantic points:

      -The authors are Denon Start, Ben Gilbert, and Andy Sih, not just Start & Gilbert. My presumption is that the first author was primarily responsible for designing the study, collecting the data, and writing the paper, but I obviously have no firsthand knowledge as to how authorial responsibilities were divvied up.

      -As I said in the post, I do think there’s a defensible scientific rationale for the non-random assignment of organisms to treatments here. It’s more or less the same rationale as in, say, Dolph Schluter’s 1994 character displacement experiment on sticklebacks. You do some sort of manipulation so as to ensure that there’s a lot of phenotypic variation for natural selection to act on. Sometimes, a good way to get information about how “realistic” or “natural” systems work is to create unrealistic, unnatural conditions. As I said, I’m not clear why that rationale wasn’t stated in the paper originally. I agree with you that it doesn’t seem like the sort of misstatement that would happen by accident. And I agree with you that it’s totally inappropriate to misstate one’s methods. Whether it’s the sort of misstatement that authors should be allowed to correct rather than having to retract…I have my own views, but I suspect this is a point on which people will disagree.

      -Yes, “my laptop was stolen” is the stereotypical excuse for not being able to provide raw data. I don’t have any first hand knowledge, but my possibly-wrong impression from what I’ve heard, and read on social media, is that the first author’s laptop really was stolen. Which of course doesn’t explain or excuse why the data weren’t backed up. It’s obviously bad practice not to have multiple backups of your data, especially in this day and age of free and easy-to-use cloud storage.

      Re: retractions for not sharing data, Am Nat (where I’m on the editorial board) has just adopted a policy that gives the journal the option of retracting papers that don’t comply with the journal’s mandatory data sharing policy (which we’ve had since 2011 if memory serves). Going forward, I think it will be interesting to see if Am Nat and other journals do start retracting papers for failure to comply with data sharing policies. And if they’ll put procedures in place that will prevent papers from being published in the first place unless they comply with journal data sharing policies.


      • Oh, I’m with you that not all ecological science need be random. In my work, pollution effects in streams, we don’t sample randomly. We target. Above and below a mine, or outfall, or tributary, or select for or against certain channel or other features, such as proximity to hostile landowners or passerbys. But saying it was random when it was deliberately non-random can not be explained as an oversight. I was overly snarky in my comment about the data vanishing with the stolen computer, but if the data were actually deposited as stated they would have had a backup. Few scientists get training in data management and many don’t know/follow best practices, but this was pretty bad.


        What I was getting at with the thought of a non-judgmental retraction that allows authors to republish elsewhere is that, as in everyday laws, their should be graduated penalties for problems in published articles:


        Good citizens cleaning up after themselves: Simple corrections for minor oops equate to parking tickets (µg/L transmogrified to mg/L or table rows and columns misaligned during typesetting, minor calculation mistakes, …);


        Misdemeanor violations: embellishing or obscuring details to make a generally solid study sound better than it is, …). That’s where Start and Gilbert are. There should be proportionate penalty, like having to go through the publication process again with a revised honest version, in a less rarified journal than Ecology Letters (https://dx.doi.org/10.4033/iee.2012.5.4.e)


        Felony violations: Dry labbing, egregious image falsification, mistreatment of humans, … Penalty should be retractions and career change.

      • I want to step away from the specifics of the Start & Gilbert retraction and start a broader convo about the idea of “levels” of consequences for different sorts of problems in published articles.

        I’m intrigued by the idea of a “retraction without prejudice”, analogous to how some journals (Am Nat is one) have a publication decision called “decline without prejudice. “Decline without prejudice” means, roughly, “We think this ms might contain the core of a publishable paper. But the revisions required to meet the publication standard would be substantial, and we’re not sure they’re even possible. So we’re declining the ms, but open to considering a heavily revised resubmission in future.” “Retraction without prejudice” could be sort of similar. As you say, the ms is retracted, but the authors are welcome to revise the ms appropriately and resubmit it (I’d suggest they be allowed to resubmit to the previous journal if they choose.) As opposed to “retraction with prejudice”, which would mean “this ms is retracted for unfixable problems. We will not consider any resubmission no matter how heavily revised, and no other journal should either.”

        I know that there are other areas of life that have something similar, but I know little about them. In some jurisdictions, can’t lawsuits be dismissed, or else dismissed “with prejudice”, the latter meaning that the lawsuit can’t be refiled?

        On the other hand, one could argue that we already have a sufficient number of “levels” of remedies for dealing with published papers that have problems. We already have correction, expression of concern, and retraction. We don’t need multiple grades of retraction.

        One could also argue that detailed retraction and correction notices remove at least some of the need for further, more finely-graded “levels” of correction and/or retraction. Readers can read the correction or retraction notice, and decide for themselves which “level” it occupies in their own minds.

        Part of the issue here is that the problems we want to address with corrections/EoCs/retractions vary on multiple dimensions. They don’t just vary on a single dimension of “seriousness”. For instance, “was the problem accidental or intentional” is a different dimension than “to what extent does the problem undermine the conclusions of the paper”. Returning to the Start and Gilbert correction: the misstatement regarding non-random assignment of organisms to experimental units strikes me as a problem that was surely intentional (as you noted). Maybe I just lack imagination, but I find it hard to imagine how an author could accidentally misdescribe that method of assigning organisms to experimental units as “random”. And I think such intentional misstatements of methods are totally inappropriate. But it also strikes me as a problem that, on its own, doesn’t much affect the paper’s conclusions. So, a serious problem on one dimension, not a very serious problem on another dimension. What’s the appropriate remedy for such cases? I have my own views (which I’m choosing to keep to myself), but this seems to be a case on which people disagree. The journal EiC obviously felt a correction was the appropriate remedy, but I’ve seen comments from others arguing that the paper should’ve been retracted.

  2. Maybe it is just me but I would suggest perhaps changing the double quotes to single quotes in your summary. You were pretty clear with your “in so many words” but I quickly read it and thought this is what the authors had literally written (which is much more damning than what they actually wrote).

  3. Pingback: Friday links: Epstein fallout continues at Harvard, memes vs. intro biostats, and more (includes quick poll) | Dynamic Ecology

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 )

Google photo

You are commenting using your Google 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 )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.