Retraction Interaction

During my graduate studies, I've come to appreciate that there are several problematic areas of scientific research and academia. The peer-review process, the cyclic nature of funding begetting publishing begetting more funding, sexism and racism within STEM fields, p-value fishing, and more recently, retractions.

Several articles in the past few months have delved into the rise in retractions of scientific papers, and the varied effects it can have within the scientific community. I highly recommend Jill Neimark's piece in Aeon Magazine from December: "The Retraction Wars - Scientists seek demigod status, journals want blockbuster results, and retractions are on the rise: is science broken?" Talk about a fantastic byline, and the article truly delivers a comprehensive look at what retractions are, how they're on the rise, and why they're both necessary and detrimental to scientific advancement. But what are they?


Neimark gives a very satisfying definition of a retraction [emphasis mine]:
The edifice upon which science is built is self-correct. And self-correction generally works. Scientists make mistakes, and science corrects those mistakes. This happens when results cannot be reproduced, and the original work is found in error. An erratum is issued when errors are relatively minor, and do not invalidate the basic assumptions and conclusions of the study as a whole. A retraction is issued when the study is no longer valid. A retraction withdraws, refutes, or reverses the entire scientific finding. 

In October, Nature reported on the rise of retractions - despite the number of publications increasing over 40% in the last ten years, retractions have increased ten-fold. 44% of those retractions are from instances of scientific misconduct, such as data fabrication or plagiarism. These results were derived from Web of Science and PubMed notices - both major journal databases that researchers often use to find scholarly articles. There isn't a reliable database that tracks retractions, instead, you must search through databases like Web of Science for retractions or look-out for retractions from individual journals. Or, you can follow blogs like Retraction Watch, which mainly report on retractions within life sciences research.


Why commit misconduct?

There are several reasons for scientific misconduct to occur, but it mainly stems from the need for scientists to publish in peer-reviewed journals. Publish or perish, as the phrase goes. Doctorate students need publications for their dissertation, post-docs are expected to publish to keep their jobs, and major researchers require publications in big-name journals in order to maintain their posts and status, as well as receive grants.

But when peer-reviewed journals operate as a business that need to maintain prestige and impact factor (the measure of how important an article published within a journal is), journals cannot afford to publish every single study that gets submitted. Even open-access, online-only journals such as those in the PLoS publishing family will curate admissions to their publication standards. There is also a bias among scientific journals to only publish significant and novel studies - if your experiment yielded insignificant results or is a repeat of a previous study with supporting results, it is unlikely that it will be published. This, in addition to the time required to peer-review articles, adds to a mentality of competition among scientists for publication. This in turn leads to errors, data fabrication, and lack of disclosure of funding sources.


The infamous Wakefield case

This internal toxicity has effects on society, including the level of 'trust' in science. Climate change deniers, opponents of child vaccinations - all find support in both articles that make it past the scrutiny of peer-review as well as the retractions of major research. The most famous example of a retracted paper is that of Andrew Wakefield's research on MMR vaccinations and autism spectrum disorder. The most famous study was published in The Lancet, the premiere journal of medical research.

Published in 1998, "Illeal lymphoid nodular hyperplasia, non-specific colitis, and pervasive developmental disorder in children" was used to support the theory that an MMR vaccination can induce autism in children. Heralded as an advancement in medicine in the late 1950's, vaccinations turned into a sort of boogieman for concerned yet misinformed parents (and still are to this day). British journalist Brian Deer is credited for taking down Wakefield, as he investigated the study and reported his findings throughout the mid-2000's. This came to a head with a General Medical Council panel on the conduct of then Dr. Wakefield. Wakefield was discredited and stripped of his UK medical license. The study was retracted from The Lancet shortly after, plenty of time for the study's bogus results to permeate public discourse.

via Skeptoid

Since then, Wakefield has been a martyr to the anti-vaccination cause; when you Google "Andrew Wakefield," half of the hits returned are from anti-vaccination and conspiracy blogs. Despite the fact that his study and entire career was discredited and proven to be falsified, his work is still used as scientific support within certain circles. The routine retraction was intended stop misinformation from spreading, but instead lends an air of conspiracy to the idea that vaccinations are unsafe or a huge pharmaceutical scam.


The repercussions of retractions

The work of Andrew Wakefield is one example of scientific misconduct and subsequent retraction. More recent cases include those of Haruko Obokata, a stem-cell researcher who has been in the news of late about fabricated data in her latest study. Obokata and co-authors claimed to be able to induce pluripotency of cells with a mild acid bath, thus 'creating' stem cells. Stem cells are used in a variety of medical studies and are being considered for therapeutic treatment of several human degenerative conditions, but they are expensive to source and embryonic sources raise ethical concerns. This would be a major breakthrough for stem cell research - unfortunately, it was too good to be true.

Replication in the lab was impossible, and investigations into the work found duplicative publications and misleading figure captions. Nature, who had published the work and an accompanying study in January 2014, swiftly retracted both papers seven months later. This case of retraction is tainted by tragedy. A month following the retraction, supervisor and co-author Yoshiki Sasai was found dead in his lab. He had killed himself. While he had been cleared of scientific misconduct, Obokata was Sasai's protégé, and he had faced criticism regarding lack of oversight, citing in his suicide note that the media frenzy and shame was too much to bear.


Retractions in ecology

So how does my own field fare in terms of retractions? Retraction Watch mainly works in life sciences, though one search of "ecology" yielded Jesús A. Lemus, whom I'll speak on later. A search on the Web of Science using "ecology" and "retraction" wasn't that useful, returning two useable articles. While figuring out the perfect search algorithm in Web of Science used to be my favourite past time (lie), I decided to go with a different approach.

First, I focused on the PLoS family of publications, if only because their user interface is for more accessible than other journals (I'm looking at you, Elsevier) and they are open-access, so anyone can perform this search. If you search for "retraction," you get 101 results, though not all are retractions. I found quite a few that I consider to be within ecology or conservation biology, and they fall within different types of retractions, from honest mistakes to full-blown scientific misconduct.

Honest mistakes and unreliable data


These cannot be considered scientific misconduct, and as such the authors should not be vilified or considered untrustworthy. Often, the author will seek to find a solution to the problem, and barring a solution, will offer retraction. The internal struggle between author and editor may be a bit more complicated, but in the end, no one is really hurt or penalized.

Mann et al 2012 "Multi-scale inference of interaction rules in animal groups using Bayesian model selection" This retraction clearly states that the authors are retracting the paper. Richard Mann, lead author, explains that there was a bug in the Matlab code that ran the statistical analysis, which severely limited the sample size used in analysis. An unfortunate mistake, but an honest one. Mann intends to reassess the data, bug-free, and seek republication. And he did! So, good on you, Dr. Mann.

Kamilar and Marshack 2011 (in Biology Letters), "Variation in chimpanzee 'culture' is predicted by local ecology, not geography" This is another example of errors in the dataset.

Georgette 2012 "Predicting the Herd Immunity Threshold during an outbreak: a recursive approach" An author-initiated retraction due to a mathematical flaw regarding model assumptions
"P-values" from xkcd
Plagiarism and self-plagiarism

Liu and Xue 2012 "Functional conservation of the Drosophila gooseberry gene and its evolutionary alleles" Outside parties brought concerns to PLoS ONE editors about this data, and the original authors did not have access to the raw data. This is in addition to some self-plagiarism concerns as well as inadequate inclusion of funding bodies. Therefore, the article was retracted. While this doesn't look good on the part of the researchers, it is definitely not the worst case of self-plagiarism out there, and the comments on the retraction make it look like more of a lab dispute than complete wrongdoing.

Starkey, Smith, and Barnard 2011 "Fisheries and marine animal populations: Learning from the long-term" A case of plagiarism (a "substantial part of the text" in fact) wherein the lead author takes responsibility.

Fabricated data and other concerns

Researcher Dr. Jesús A. Lemus has racked up a total of 12 retractions so far (according to Retraction Watch) following an ethics investigation by the Spanish Superior Council of Scientific Research (CSIC). His work mainly deals with wildlife and livestock disease. In addition to allegations from CSIC of scientific misconduct, Lemus may have invented one of his co-authors on the most recently retracted paper in Environmental Research. He has had an additional four retractions in PLoS ONE.

The final example I have reads like a case of academic rivalry and/or lack of professionalism. Anders Pape Møller is a huge figure in evolutionary biology. My colleagues may recognize the name (here's something to jog your memory: the length and symmetry of tails in barn swallows). The details of the case can be read here. Essentially, fellow evolutionary biologist Jørgen Rabøl wrote an opinionated response to Møller's Oikos paper on leaf symmetry, stating that the data must be fabricated. Unsatisfied with a retraction by Oikos, Rabøl took the case to the Danish Committee on Scientific Dishonesty. In rebuttal, Møller states that this is a case of a petty grudge and planned to sue Rabøl (there is no follow-up on the issue, and both men are still employed at their universities, so something must have been worked out in private).


Conclusion: Scrutinize all the things

Retractions are necessary to science, but you cannot ignore that if used as a weapon or fodder for news outlets, they threaten the professional and personal livelihood of authors. Instead, we need to refocus on why retractions happen in the first place. To err is human, but at what point is it an error due to chance or circumstance, or an error due to publishing pressure. And when it's scientific misconduct, the 'how' it occurs is just as important as the 'why' it occurs.

Finally, things don't seem to be so bad in ecology and conservation biology; aside from a few rotten apples, we seem to have a lower amount of retractions than say, medicine. But then I find myself asking, is that because we're just generally better at catching mistakes before publication, or do they slip past easier because our research isn't being scrutinized as closely as it should be? When the results of population studies or ecotoxicology assessments are presented to government bodies for inclusion into policy planning, we need to be sure that this is a study can withstand the scrutiny of opponents. And that means more replication and follow-up research alongside those novelty, next-generation studies that continue to grace the covers of Nature and PLoS ONE.



via Nature
And for those of you in the lab or field: if you suspect a colleague of scientific misconduct, know how to report it to the proper authorities. Aside from the somewhat lofty ideals of protecting the sanctity of science, there are the personal repercussions as well. You may want to distance yourself from the act - I didn't do anything wrong, I'll just stay quiet and things will blow over - but if and when the shit hits the fan, your work's integrity may be questioned as well. In this case, one rotten apple can sour the barrel that is your research group.
 

-Kim

--------------

If you're interested in more retraction scandal, I suggest looking into social psychologist Diederik Stapel (54 retractions so far!*), anthropologist Mart Bax, or economist Peter Nijkamp: all are relatively influential in their field, and all are Dutch...


(11.24.2015)
Update: Retraction Watch has announced (with their parent network, Center for Scientific Integrity) a partnership with the Center for Open Science to create a retraction database. At long last, they are building the foundation for one of their major goals!
*Also, poor Diederik is now up to 57.

Comments

Popular Posts