To the practicing scientists here, especially those with editorial experience, I have a set of questions regarding about the differences between retractions and corrections.
What is the difference between a correction and a retraction? Is one more serious than the other?
What is the criteria for determining if a mistake should be handled as a correction or a retraction? Severity? Impact and citation by others? Centrality to argument of the mistake?
In the literature and elsewhere, how should a correction and retraction be handled differently? How should they be handled the same?
Is promptly addressing important to research integrity? If so, what are reasonable timelines by which to determine promptness?
In engaging the public, should our standards if error management be higher, lower, or the same as in our professional publications? Why?
I am asking these questions to calibrate my understanding to the scientific community. For this reason, I’m not interested (or permitting) digression into adjudicating any particular examples, and non scientists should restrict their comments to a comment thread which the @moderators can create if needed.
Disclaimer: I’ve always been more on the consuming rather than producing end of the literature. My thoughts below are from my observations as a scientist and science educator (we teach a course that includes research ethics) rather than any editorial or extensive authorial experience.
For me there is a difference between a correction and a retraction. In my mind a correction is for typos, editing mistakes, and easily correctable analysis errors. In other words, what was wrong is minor (not affecting the main substance of the paper) and the fix is clear.
A retraction, on the other hand, is for cases where there is an error that affects the substance of the paper (e.g. conclusions we drew were invalidated). I think importantly that retractions also tend to not have an actual “fix” at the time. I think of something like: we know that this data/analysis is wrong, but we don’t immediately know how to fix it so we want to withdraw it from the body of the literature until we can figure it out.
A correction is “I stand by the substance of what I said, but I’m only human so I’m going to try to eliminate unintentional errors.”
A retraction is “This error is significant and changes the substance of what I wrote, I can no longer stand by this statement/conclusion/paper.”
My take, and this could be different for different kinds of works (research articles vs. pop or commentary articles, etc.)
1.) Correction is appropriate when the error in question is minor and doesn’t affect the overall conclusions that the paper makes. If the “main point” is unaltered, it’s a correction. However, if the error changes, or undercuts the main support for, the main conclusions of the paper, retraction is called for. I have had to issue one erratum in my career. A graph was mislabelled giving the wrong units. (ug instead of mg) As embarrassing as it was, it didn’t change the conclusions of the paper.
2 and 3.) It depends on the outlet. (And I don’t really get what you’re asking here anyway)
4.) Of course. And Timeliness is important, but so is care. Authors should take every effort to fully understand what happened and what “the reality is” before issuing a correction or retraction. Like research itself, corrections should be carefully managed.
5.) Again, I don’t really know what you mean here. The presentation of original research results is just a totally different thing than writing about things for the public, it’s apples and oranges and I think a comparison of “the standards” just doesn’t make a lot of sense. Here’s why I said that…
Public engagement is kind of like the discussion section of a research article. It’s the place where interpretation and pontification are not just allowed but expected, so there is a lot more leeway for opinion and that should be understood by all, including the audience. If someone writes an opinion piece in a blog or even The NY Times about a hot research result and then, as the field evolves over the next ten years, the result itself is seen in a very different light, the opinion piece may be rendered obsolete or even flat out wrong. No correct is needed in that case. It was understood to be an opinion at the time. In fact, if we read the science pages of the NY times from, say, the early 90s, we will find tons of articles that now don’t make a lot of sense considering everything we learned since then. Should we correct and retract them all? Of course not. If we did, we’d spend more time looking backward at things we’ve written decades before than looking forward toward new projects. If something we write is wrong or misleading at the time we publish it, yes, a correction or retraction is warranted, but if new information later renders it obsolete or wrong, I don’t think there is any obligation to correct the record. I think opinion, commentary, and “Sci Comm” in general is understood to be an “in the moment” commentary of the state of the field with a very short expiration date. What’s the timeline? Who’s to say?
I will answer based on my specific experience at Cell Reports, a primary research journal. I will not claim to speak for all primary research journals, but I think you can assume that my positions are in fact representative of journals in the life sciences and probably of journals in all of the hard sciences. I consider it obvious that the scientific peer-reviewed literature is substantially different from other kinds of writing and literature.
Many of the answers to these questions can be found in the work of COPE (Committee on Publication Ethics), of which Cell Press (and hundreds/thousands of publishers and societies) is a member.
There is a big difference. Corrections leave the literature intact, and are made with the premise that the uncorrected article was still valid overall. The Correction text identifies the error and how it came about, then provides the corrected text/data. The Correction is linked to the original article explicitly.
Retractions remove articles from the literature. They do this openly. I think everyone finds a retraction to be “more serious” than a correction.
Between a straightforward correction (say, one that corrects a substantive error in a figure or table) and an all-out retraction (say, one that removes a fraudulent article from the literature) there lies a range of errors and problems that require subjective judgment. The main criterion is centrality of the error to the argument in the paper, which is related to “severity” I guess. Other criteria include more subjective judgments about overall trustworthiness of the work, and let’s be open here: errors that suggest negligence or incompetence can and should be judged differently from what we should call honest mistakes. There are no checklists or rubrics by which these judgments are made, but there are guidelines established with great care by COPE; when I face situations like this, I confer with colleagues and sometimes convene a committee of other editors to consider and judge the response to a paper containing errors.
See above for how they are different. I don’t know what anyone means by “handled” but both corrections and retractions involve these two questions: 1) is there an error that needs to be dealt with? 2) If so, how should it be dealt with?
Yes, of course promptness is important. In some cases (cf. the bogus “data” put on bioRxiv about the Wuhan coronavirus), “prompt” means days. In other cases, there might be less urgency. It depends on the error, but in these days of online publication, there are few good reasons to delay a correction. But the truly complicated scenarios all concern retraction and not correction. If a journal suspects a paper to be unreliable, and is considering retraction, it is obligated to engage a process that is nearly impossible to do in a matter of days or weeks. For this reason, something called an “editorial expression of concern” can be used: the journal attaches this notice to the paper, announcing that the journal has some concerns that are serious enough to warrant scrutiny (i.e. an investigation) but which are unresolved. Nevertheless (and this has not yet happened to me as an editor-in-chief), if/when there is clear urgency (an error that endangers people, for example), the process can and should be expedited.
First, I know perfectly well that you are building a garden path on which to bash BioLogos, and I think all readers of this conversation should keep that in mind. I considered deleting all of my responses at this point, because I’m unwilling to be a part of your cynical and selfish game. I’m tagging @glipsnort here specifically, not to solicit a response but to signal my discomfort with your behavior and to make sure at least one other participant at BioLogos sees this.
I don’t think there should be a different standard for correction when talking to the general public, but I also don’t think that the challenges of communicating science in 2020 have anything at all to do with corrections or retractions. Consider a case in point, which I think illustrated exemplary conduct by a journal (my friends and colleagues at Current Biology). [Disclosure: these are indeed both friends and colleagues, which should be noted as readers consider my words.] The journal published a paper on effects of religion on altruism. Then, months later, they published a reanalysis of that paper, by different authors, which invalidated some of the findings of the original paper. The verdict was clear: the main conclusion of the original paper, which garnered major news coverage, was wrong. The data were fine, but the conclusion was wrong. The original paper was retracted.
Here’s why the story is relevant here. Science worked just as it should. The original authors made an error in their analysis. A second group requested, and received, the raw data, reanalyzed it, and demonstrated the error. The paper showing the error was published in the same journal as the original paper. All good, right? Wrong. In the popular media, there was no correction or retraction. Years after the result was known to be wrong, it was being cited in blogs and news pieces. Science did its part to correct the mistake, but the cat was out of the bag–more accurately, a millions cats were out of the bag, and there is just no way to round them up. My point being: you are deluded if you think that scientific processes of error correction–retractions and corrections in the literature–can adequately influence truth and error in a population unequipped to detect even the most blatant of lies.
But with all that said (and shit, sorry this is way too long), when I blogged regularly, I did corrections, and tagged them as such, and even invited “peer review” of a few of my posts. Maybe because I’m a scientist who has been immersed in the literature for more than 3 decades, I was unwilling to delete posts entirely once they had been published and certainly not after they had been commented on or linked to by others. Blogs, of course, are easy to revise. If I ran a website that published articles on science, I would on principle never delete an article and would instead follow processes akin to correction and retraction at journals like Cell Reports.
I’m trying to figure out why you needed to ask about the difference between retraction and correction. That’s like asking about the difference between a research article and a book review. Looks like a garden path to me.
Well, to be honest, you recently and correctly called me out on a mistake of calling something a retraction that wasn’t. You were right, and I immediately corrected it.
I wasn’t speaking in hyperbole at the time, but made a real mistake, and significant one at that. This provoked some introspection, and I wanted to be sure I understood this well so I did not make a related mistake again in the future.
Turns out, I agree with your answer but also wouldn’t have said it with such clarity. I supposed I soaked up culturally the meaning, but did not have some of the details in place, probably because I’ve never been responsible for a large journal, like you. As one example, I had never heard of COPE, and never read those guidelines. I would still not know about them had I not asked you.
As for “garden path”, I’m not sure what you mean. As I’ve noted to you privately, I did not plan to (and can publicly state I won’t) quote or cite your comments in this thread in any context. If it makes you more comfortable, I can even make this a private thread.
I’m glad you asked about retraction and correction. I would not have assumed that a busy scientist would have read the COPE documents, but I did think you should understand retraction and correction. It’s a good sign, for you, that you haven’t had any personal experience with these things.
There is no need to make this private and you are welcome to cite my comments about retraction/correction and how journals think about it.
To understand why I smelled a garden path, consider the context of your harshness toward BL people, and look again at your last question. Yes of course the questions look fine outside of any context, but that’s not where they live. Food for thought.
I applaud that. BioLogos bears that responsibility because it has earned respect. Other organizations that frequent your forum do not. Let’s be frank: you aren’t going to subject the pitiful Discovery Institute to this kind of scrutiny, are you? Do you plan a series of posts calling on those authors and that organization to correct errors or retract falsehoods? Again, I agree that BL should be held to a higher standard, but that’s because it deserves to.
Well, yes I did, and I do. So I’m not sure what you are getting at here. In fact, I have regularly stated that ID has major problem with trust because they won’t retract anything. Though occasionally and rarely there are exceptions by one or two scholars, most of them never do. It is anti-scientific.
I pressed the case on ID scholars quite often. In the case of Behe, they were pretty unhappy with me, and he still did not retract (and is part of why we don’t see @Agauger here any more). I’m not exactly sure how you’ve come to the impression I use kid gloves on them.
The difference between DI and mainstream scientists is that I don’t expect DI to do right on these points, at least not for the most part, but do expect mainstream scientists to do right. Moreover, I don’t have credibility to call DI to the mat (as I often do) unless I’m willing to do the same to others in mainstream science.
I also agree with this. Honestly, my problem is not with BioLogos per se, or with the majority of scientist affiliated with them. It is with some (in my view) unscientific arguments and unscientific ways of managing errors that is hopefully confined to a few people and on the topic of population genetics. I suppose one reason I tend to react so strongly, on a personal level, is because it reminds of me of the dishonest versions of creationism (in this confined area, to be clear). That is not something I personally know how to tolerate from mainstream scientists yet.
Good to hear. Of your two examples, both are Behe, whose abuses of science are petty compared to other crap, and one is about jacket blurbs. I haven’t seen calls for retraction or correction, but if you regularly do this, then great.
Both Ann and Paul Nelson are tragedies IMO. Both should know better. Really sad.
As for “kid gloves,” I don’t mean to say that you don’t criticize the pitiful crap that the DI coughs up. I’m just uncomfortable with a tone used against BL that seems out of proportion. But look. I’ve had my say. My parting word is this: I do think that deleting an article, about science, from a website of the prominence of BL’s, is a mistake. My preference is what I would have done: leave the post there, with clear notes that it has been corrected, pointing to the post where that happens. That approach would be consistent with the practices of the professional scientific literature, and it would be the opposite of what we all expect from anti-science disinformation dispensaries.
I’m actually not sure that the norms of scientific publishing provide the best framework for this issue. Retractions and corrections in the scientific literature normally concern one’s own findings and methods in primary research. I don’t recall seeing a correction to a paper (much less a retraction) because an author misinterpreted some aspect of their field in a paper’s introduction or discussion. Instead, mistaken ideas are generally refined or replaced in new publications, by the same or by other authors. I suspect broader principles of integrity, accuracy, and transparency provide a better basis for framing this discussion.
I agree. I think it is a shame that @Agauger left here, but its got be hard when she does seem to follow a different path than the rest.
I do this regularly, and also on far more tehcnical points. I picked those because they are understandable to the public, which is critical if you are going to call out a scientist on something he is unlikely to correct. You need to do that on things the public understands. If he won’t retract things everyone knows he got wrong (and are not central to his case), then we can’t trust him to retract more nuanced things.
Of course, I’ve addressed many more of the technical points in detail. In those cases, my goal is helping the public understand, and also ensuring I didn’t miss something. See this for example: Which Irreducible Complexity?.
I do think my tone can be different with BioLogos, largely this is when I’m dumbfounded that they are not following standards I just expect everyone to follow in mainstream science. I’m not surprised, in contrast, when a young earth creationist scientist executes nonsense. That’s what I expect from them.
Well, now that we are talking about it, that is one of my concerns. Though I’m sure BioLogos is not following the COPE guidelines, Haarsma did articulate a policy, a policy they seem to have violated the same day she announced it.
I agree with this too. However, that silent deletion fails on these grounds too.
This is where I also feel there is a massive missed opportunity. Handling these mistakes well would have put others (e.g. DI) really to shame, and would take away any reason I personally have for objecting to them. We all make mistakes. Some of us make large mistakes. And sometimes we don’t fix them quickly enough. If we put that all out there, and are transparent and apologetic, well then I think everyone should move to asking why other origins organizations are not doing the same. It is fairly tragic that this is not what is happening now.
To be clear though @glipsnort, Dennis made several novel claims that are not found in the mainstream scientific literature. I don’t know how closely you’ve examined his work, but many of his primarily claims and conclusions were never established findings in the literature, and in fact were never subject to peer review. That is fine of course, except they were presented as the established consensus of mainstream science, and no scientist at BioLogos was able, willing, or allowed to question his scientific work.
Sorry for the side note, but this one made me LOL because if they were to start retracting errors in their articles, the effort would completely consume them for years. Every day brings new articles with factual errors in them.
Both are serious, but correction is less serious than retraction. Correction means you described something incorrectly, or there was some error in the methods or data that were inconsequential to the conclusion. A retraction usually means that there were major flaws that invalidate the conclusions.
There is a fine balance between punishing people for publishing poor papers and encouraging scientists to be honest in reporting errors. I think we need a little more carrot and a little less stick.
Promptness is part of being an honest scientist. Report it as soon as you learn of it.
I think there is also an expiration date. I don’t think scientists should be going back to 30 year old papers and fixing them in the light of new technologies or newly understood errors inherent in older methods. A new paper is probably called for at that point. It’s a tough call, though.
Standards should be just as high when we communicate with the public as they are when we communicate with other scientists.
One of the hurdles is the amount of nuance you are able to use in each setting. Peer reviewed papers use strongly hedged language and are full of nuance with respect to methods and data. That language is very confusing to the public, and even to scientists from different disciplines. Communicating to the public requires a balance of specificity and generalization, and error can become a problem if those are balanced incorrectly.
Below are 5 examples, 4 corrections and a retraction, from the latest Nature.com Newsletter. The retraction appears to be a case of rush to publish before we get scooped, and retracting when they could not verify their own results. This is unfortunate, but not uncommon is some very competitive fields.
I see hundreds of manuscripts in the last 20 years, and therefore a lot of corrections that slipped pasted the first reviews. I don’t recall there ever being a need for corrections due to statistical errors. There could be minor corrections I was never made aware of.
I don’t have any papers that were retracted. I’ve seen a few where the statistician (myself or others) refused to participate unless changes were made. I don’t think I have any published papers that are out-and-out wrong - but there are a few that might be useless.