What is the Goal of Retraction?
By Marianne Evola
A couple months ago, I wrote about my mixed feelings regarding the Retraction Watch website. For those unfamiliar with the site, Retraction Watch is a website that tracks and reports all of the publication retractions from the research literature. They investigate and report the reasons for retraction, to the best of their ability. In early January, the site engaged in a discussion that returned my thoughts to the role of publication retraction and of the Retraction Watch website. Specifically, the report that caught my attention was Why retraction shouldn't always be the end of the story. This report described an assertion made by a few researchers that retraction resulting from misconduct should clearly indicate and describe the misconduct that led to retraction. Too often, they report, retractions imply that honest error led to retraction rather than discovered misconduct. In other words, while retracting, the authors failed to disclose that an allegation or finding of misconduct triggered the retraction. This assertion raised several red flags in my mind, regarding the role that retraction plays in correcting the scientific literature. Specifically, if retraction becomes a tool for punishment, researchers may be reluctant to utilize retraction to correct their errors in the literature. Ultimately, this would damage the research record and impair research productivity.
One of my early lessons in graduate school was that mentors needed to be approachable if they wanted to maintain the integrity of data. Initially, approachability seems like an easy goal. Most of us were taught lessons about kindness and respect as children. However, being an approachable mentor means that you have to be approachable when students or subordinates fail to follow basic instructions. Their failures can seriously mess up important projects and waste considerable time and expensive resources. Students have to be willing to confess their failures to their mentor and be confident that their mentor will work with them to correct the problems, or in the worst cases, discard the failed research and resign themselves to the loss of resources. If students or subordinates feel that their failure(s) will be met with shame, humiliation and punishment, they would be more likely to hide their failure, which could impair the integrity of the research. In truth, after years of mentoring students, I learned that seldom was there a need to shame or humiliate a student for their failures, even the avoidable errors. Generally, students were already sufficiently embarrassed when they came to me with the problem. Thus, the model of the approachable mentor is an efficient means to protect the integrity of research. It ensures that errors are acknowledged, poor research practice is corrected, effected data is appropriately labeled for omission and a plan is constructed so that the student can recover his status in the lab.
Publication retraction is a similar system of correction and recovery for published researchers. Historically, the system of retraction has been a tool that contributes to the self-correction of science and the scientific literature. Retraction enables researchers to correct experimental or interpretive errors that they have contributed to the literature. Appropriate use of retraction protects our colleagues from wasting time and resources by attempting to replicate or utilize flawed results and conclusions. Ideally, retraction due to honest error should not damage the reputation of the scientists that appropriately decide to retract flawed research. Thus, retraction should not be punitive. A system of retraction that is not punitive empowers authors to correct their errors, just as approachability enabled my students to report their failures. In reality, although retraction due to honest error is not punished, it can be embarrassing (again, similar to my students). No researcher likes to believe that their work could progress to completion, through peer review and publication with a major flaw that requires retraction, because most researchers strive for the unrealistic goal of perfection. That being said, ethical researchers would choose to retract errors rather than purposely contribute to flaws in the scientific literature.
The system of retraction is a tool that contributes correction to the literature but like most tools, it's efficacy is limited. One of the weaknesses of retraction is informing colleagues that an article has been retracted. Frustratingly, long after an article is retracted, colleagues continue to cite the retracted research. Even with convenient electronic access to journals, citation of retracted research continues to be a problem (Collins, 2015). Most of our colleagues access literature online where retraction would be indicated. However, for those of us that still read best on paper, printed and filed manuscripts are still a primary sources from the literature. Once we print an article, it is less likely that we will return to the electronic source where a retraction may be indicated. In addition, some of our senior colleagues still use printed journals on library shelves as their primary literature sources. To exacerbate our love of paper, there can also be very long delays between publication and retraction, even in cases where research has been identified as fraudulent. A recent example was a six year delay for a paper to be retracted after it was deemed fraudulent and retracted by the home institution of a researcher found guilty of misconduct. Therefore, even when used appropriately to correct the literature, the correcting impact of retraction is limited.
Similar to my views on the appropriate use of retraction, Retraction Watch celebrates investigators that actively and appropriately sought to remove their errors from the literature. On the Retraction Watch website they have an archive for authors "Doing the Right Thing" and they list authors that have utilized retraction to correct their errors. Rewarding and celebrating authors for publically addressing their errors should be encouraged. Teachers and parents across the world are aware of the impact of basic behavioral reinforcement and they know that rewarding good behavior will increase the probability that the behavior will be repeated. If we want to improve research, we need to follow that example and reward and celebrate good research behavior. Celebrating the appropriate use of retraction to correct the research record will encourage investigators to do so, even if it is initially embarrassing to admit their mistakes.
However, increasingly, retraction is becoming associated with misconduct. Although retraction is a combination of retraction due to honest error and retraction due to research misconduct, discouragingly, misconduct seems to account for most publication retractions (Fang et al., 2012). In effect, publication retraction is inadvertently becoming linked with misconduct. The link with misconduct makes the retraction process punitive because retraction causes colleagues to question the integrity of researchers, even if they have appropriately utilized the tool to correct their mistakes. And, as stated above, a punitive process will discourage investigators from proactively retracting research and helping to correct the literature. Therefore, utilizing retraction as a means to punish or pass judgement on colleagues that have been found guilty of misconduct would only serve to make retraction synonymous with misconduct.
Investigating and punishing misconduct is the role of an investigator's home institution or the role of government research sponsors. Investigation committees appropriately take great effort to protect confidentiality of all parties (i.e., complainants & respondents) to ensure that everyone's reputation is protected until such time as it has been clearly demonstrated that someone did irreparable damage to their own career through engaging in misconduct. Furthermore, it can take years for an investigation to be completed and a verdict issued and they can consist of multiple appeals, evolving committees and growing chains of evidence. If retraction becomes merely a sanction of misconduct, investigation committees will only be able to issue the sanction once the final verdict has been made. There could be a delay of years between publication and retraction which will permit poor research to continue to impact scientific inquiry. Resources and time will be wasted and good junior scientists could have permanent damage to their careers due to wasted time.
If the goal of retraction is to correct the literature in a timely manner, retraction should not be an avenue to appropriately or inappropriately pass judgement on our colleagues. Rather, retraction should be a natural and neutral part of making a research contribution. If we are going to assert that science is self-correcting then we need to provide our colleagues with tools to contribute to the correcting process without allowing the tools to damage their reputations.
Marianne Evola is senior administrator in the Responsible Research area of the Office of the Vice President for Research. She is a monthly contributor to Scholarly Messenger.