It’s perhaps one of the juiciest words in scientific language. One might think of it as the Scarlet Letter of science, a public stamp that can trigger gossip, speculation, whispers… Something went wrong somewhere, and someone is paying for it by labeling their published work as invalid.
But what exactly has to go wrong to call for a retraction? This is the question posed at Retraction Watch. DrugMonkey is polling readers on their interpretation of the retraction and posts his stance:
… we need to be exceptionally clear in the business of science that a failure to replicate is not, in fact, evidence of fraud.
In my view, a correction suffices… in most cases where there is not fraud.
Retraction, to me, implies that there is reasonable evidence of some sort of shenanigans.
I’m certain we could all agree that shenanigans (e.g. data fraud or fabrication) should result in retraction.
I also strongly agree with DrugMonkey’s first point, that lack of reproducibility is not in and of itself a strong enough reason to retract a paper. I think anyone who’s been in the lab for even a year can understand the difficulty of repeating a protocol from someone else, step by step, in the same way and getting exactly the same answer-especially with biological systems. Sometimes there are small differences in how we do certain steps that we would never think to write down. Occasionally something as innocuous as cell passage, density, or reagent batches can affect our results. Environmental changes (e.g. building temperature or humidity) can wreak havoc on instruments or growth conditions.
However, with the current publishing structure, I think an inherent problem with the data does warrant retraction. I’m talking about things that directly influence outcomes and conclusions, such as contamination of a sample or reagent, realizing that your reagent isn’t what you thought or has an altered specificity, etc. This retraction in PNAS provides a good example.* Experiments were done, conclusions made, paper accepted. Then the authors realized they had “made an error in interpretation”. They go on to explain in detail how this occurred. The original paper is still accessible, now with a note to See Retraction in bold red letters at the top of the page and a link to the retraction in the side bar. This approach allows us to learn from other’s mistakes but makes it clear that the original results do not mean what we first thought.
One day the scientific community will trade the static print-type approach of publishing for a dynamic, adaptive model of communication. Imagine a manuscript as a living document, one perhaps where all raw data would be available, others could post their attempts to reproduce data, authors could integrate corrections or addenda. Maybe the future of scientific publishing will make retractions for technical issues obsolete.
But we’re not there yet-and there will be some time before we reach that future. In the meantime, we should be making it clear that retractions do not always mean that someone did a bad, bad thing, that sometimes it simply means someone made an honest mistake. After all, we are only human.
* Admittedly it’s a little odd that one of the authors did not sign, but according to the corresponding author, that’s because Xu was an undergrad student not involved in the erroneous portion of the project.
To prevent being investigated by the Office of Blogging Integrity, I should note that part of this post was originally written by me at DrugMonkey’s What does a retracted paper mean?, provided comment moderation did not eat it.