Retraction in Science
In science, a retraction of a published scientific article indicates that the original article should not have been published and that its data and conclusions should not be used as part of the foundation for future research. The most common reasons for the retraction of articles are scientific misconduct including plagiarism, serious errors, and duplicate/concurrent publishing (self-plagiarism). The retraction may be initiated by the editors of the journal, or by the author(s) of the papers (or their institution). A lesser withdrawal of content than a full retraction may be labelled a correction.
GUIDANCE FOR RESEARCHERS ON RETRACTIONS
Retraction is a mechanism for correcting research literature and alerting readers to publications that contain such seriously flawed or erroneous data that their findings and conclusions cannot be relied upon. Such unreliable data may result from honest error or from research misconduct.
- a) Retractions are also used to alert readers to some forms of misconduct, such as redundant publication (publishing the same data or article in more than one journal without appropriate justification, permission or cross-referencing), plagiarism or failure to disclose conflicts of interest likely to influence interpretations or recommendations.
- b) Retraction is usually reserved for publications that are so seriously flawed, for whatever reason, that their findings or conclusions should not be relied upon.
- c) Journals generally rely on research institutions to investigate allegations of serious research misconduct (such as data fabrication) since, in most cases, editors do not have access to all the evidence and journals are not resourced or constituted to conduct investigations. Editors therefore rely on institutions to inform them of the outcome of such investigations and will usually wait for investigations to be concluded before issuing a retraction; however, if editors obtain clear evidence of misconduct such as plagiarism or redundant publication, they may retract publications independently of any institutional enquiry.
The main purpose of retractions is to correct the literature and ensure its integrity rather than to punish authors who misbehave. A retraction can help reduce the number of researchers who cite an erroneous article, act on its findings or draw incorrect conclusions, such as from ‘double counting’ redundant publications in meta-analyses.
If retraction is due to the actions of some, but not all, authors of a publication, the notice of retraction should mention this; however, most editors consider that authorship entails some degree of joint responsibility for the integrity of the reported research so it is not appropriate for authors to dissociate themselves from a retracted publication even if they were not directly culpable of any misconduct.
Disputes over authorship: retractions are usually not appropriate when a change of authorship is required but there is no reason to doubt the validity of the findings.
- a) Authors may request an editor to retract an article when a dispute over authorship arises after publication; however, if there is no reason to doubt the validity of the findings or the reliability of the data, it is not appropriate to retract a publication solely on the grounds of an authorship dispute.
- b) In such cases, the editor will normally inform those involved in the dispute that he or she cannot adjudicate the matter but will be willing to publish a correction to the author or contributor lists if appropriate proof that such a change is justified can be provided by the authors or their institutions.
Partial retractions are not helpful because they make it difficult for readers to determine the status of the article and which parts may be relied upon.
Corrections: if only a small part of an article reports flawed data, especially if this is the result of genuine error, then the problem is best rectified by a correction or erratum. The term ‘erratum’ usually refers to a production error, caused by the journal, while a ‘correction’ usually refers to an author error.
In the same way, if only a small section of an article, for example a few sentences in the discussion, is affected by research misconduct such as plagiarism, the appropriate action may be to issue a correction rather than retracting the entire article which may contain sound, original data in other parts. In this case, the correction would note the fact that text was used without appropriate acknowledgement.
Expressions of concern: if conclusive evidence about the reliability of a publication cannot be obtained, an expression of concern will normally be issued, rather than retracting the publication immediately.
If more conclusive evidence about the publication’s reliability becomes available later, the expression of concern will be replaced by: a notice of retraction if the article is shown to be unreliable; or by an exonerating statement linked to the expression of concern if the article is shown to be reliable and the author has been exonerated.
Circumstances where the retraction of a publication may be necessary include, but are not limited to:
- When there is clear evidence that the reported findings are unreliable, either as a result of misconduct, such as fabrication of data, or honest error, for example. miscalculation or experimental error;
- b) If the findings have previously been published elsewhere without proper cross referencing, permission or justification, so the work constitutes redundant publication;
- c) Cases of plagiarism; or
- d) Reports of unethical research.
Circumstances where issuing an expression of concern may be necessary include, but are not limited to:
- a) Conflicting or inconclusive evidence of research misconduct;
- b) Suspicion that the findings are unreliable but the authors’ institution is unwilling or unable to investigate the case; an investigation into alleged misconduct related to the publication either has not been, or would not be, fair and impartial or conclusive; or
- d) An investigation is underway but a judgement will not be available for a considerable time.
Circumstances where issuing a correction may be necessary include, but are not limited to:
- a) A small portion of an otherwise reliable publication proves to be misleading, especially because of honest error; or
- b) The author / contributor list is incorrect, such as when a deserving author has been omitted or somebody who does not meet the criteria for authorship has been included.
Retractions and expressions of concern should:
- a) Be linked to the retracted article wherever possible, such as in all electronic versions;
- b) Clearly identify the retracted article, for example by including the title and authors in the retraction heading;
- c) Be clearly identified as a retraction, distinct from other types of correction or comment;
- d) Be published promptly to minimise harmful effects from misleading publications, such as minimising the number of researchers who cite the erroneous work, act on its findings or draw incorrect conclusions;
- e) Be freely available to all readers and not behind access barriers or available only to subscribers;
- f) State who is retracting the article; and
- g) State the reason(s) for retraction, in order to distinguish misconduct from honest error.
Other forms of scientific misconduct, image fraud (manipulation of images to distort their meaning) is of particular interest since it can frequently be detected by external parties. In 2006, the Journal of Cell Biology gained publicity for instituting tests to detect photo manipulation in papers that were being considered for publication. This was in response to the increased usage of programs by scientists such as Adobe Photoshop, which facilitate photo manipulation. Since then more publishers, including the Nature Publishing Group, have instituted similar tests and require authors to minimize and specify the extent of photo manipulation when a manuscript is submitted for publication. However there is little evidence to indicate that such tests are applied rigorously. One Nature paper published in 2009 has subsequently been reported to contain around 20 separate instances of image fraud.
Although the type of manipulation that is allowed can depend greatly on the type of experiment that is presented and also differ from one journal to another, in general the following manipulations are not allowed:
- splicing together different images to represent a single experiment
- changingbrightness and contrast of only a part of the image
- any change that conceals information, even when it is considered to be specific, which includes:
- changing brightness and contrast to leave only the most intense signal
- usingclone tools to hide information
- showing only a very small part of the photograph so that additional information is not visible.
Suppression/non-publication of data
Failure to publish or selective release of the findings of scientific studies. Such cases may not be strictly definable as scientific misconduct as the deliberate falsification of results is not present. However, in such cases the intent may nevertheless be to deliberately deceive. Studies may be suppressed or remain unpublished because the findings are perceived to undermine the commercial, political or other interests of the sponsoring agent or because they fail to support the ideological goals of the researcher. Examples include the failure to publish studies if they demonstrate the harm of a new drug, or truthfully publishing the benefits of a treatment while omitting harmful side-effects.
This is distinguishable from other concepts such as bad science, junk science or pseudoscience where the criticism centres on the methodology or underlying assumptions. It may be possible in some cases to use statistical methods to show that the datasets offered in relation to a given field are incomplete. However this may simply reflect the existence of real-world restrictions on researchers without justifying more sinister conclusions.
Consequences for science
The consequences of scientific fraud vary based on the severity of the fraud, the level of notice it receives, and how long it goes undetected. For cases of fabricated evidence, the consequences can be wide-ranging, with others working to confirm (or refute) the false finding, or with research agendas being distorted to address the fraudulent evidence.
Consequences for those who expose misconduct
The potentially severe consequences for individuals who are found to have engaged in misconduct also reflect on the institutions that host or employ them and also on the participants in any peer review process that has allowed the publication of questionable research. This means that a range of actors in any case may have a motivation to suppress any evidence or suggestion of misconduct. Persons who expose such cases, commonly called whistleblowers, can find themselves open to retaliation by a number of different means. These negative consequences for exposes of misconduct have driven the development of whistle blowers charters – designed to protect those who raise concerns. A whistleblower is almost always alone in his fight – his career becomes completely dependent on the decision about alleged misconduct. If the accusations prove false, his career is completely destroyed, but even in case of positive decision the career of the whistleblower can be under question: his reputation of “troublemaker” will prevent many employers from hiring him. There is no international body where a whistleblower could give his concerns. If a university fails to investigate suspected fraud or provides a fake investigation to save their reputation the whistleblower has no right of appeal.
Exposure of fraudulent data
One tool developed in 2006 by researchers in Dr. Harold Garner‘s laboratory at the University of Texas South-western Medical Centre at Dallas is Déjà vu, an open-access database containing several thousand instances of duplicate publication.
Other tools which may be used to detect fraudulent data include error analysis. Measurements generally have a small amount of error, and repeated measurements of the same item will generally result in slight differences in readings. These differences can be analyzed, and follow certain known mathematical and statistical properties. Should a set of data appear to be too faithful to the hypothesis, i.e., the amount of error that would normally be in such measurements does not appear a conclusion can be drawn that the data may have been forged? Error analysis alone is typically not sufficient to prove that data have been falsified or fabricated, but it may provide the supporting evidence necessary to confirm suspicions of misconduct.
Responsibility of authors and of co-authors
Authors and co-authors of scientific publications have a variety of responsibilities. Contravention of the rules of scientific authorship may lead to a charge of scientific misconduct. All authors, including co-authors, are expected to have made reasonable attempts to check findings submitted to academic journals for publication. Simultaneous submission of scientific findings to more than one journal or duplicate publication of findings is usually regarded as misconduct.
Responsibilities of research institutions
In general, defining whether an individual is guilty of misconduct requires a detailed investigation by the individual’s employing academic institution. Such investigations require detailed and rigorous processes and can be extremely costly. Furthermore, the more senior the individual under suspicion, the more likely it is that conflicts of interest will compromise the investigation. In many countries (with the notable exception of the United States) acquisition of funds on the basis of fraudulent data is not a legal offence and there is consequently no regulator to oversee investigations into alleged research misconduct. Universities therefore have few incentives to investigate allegations in a robust manner, or act on the findings of such investigations if they vindicate the allegation.
Responsibilities of scientific colleagues who are “bystanders”
Some academics believe that scientific colleagues who suspect scientific misconduct should consider taking informal action themselves, or reporting their concerns. This question is of great importance since much research suggests that it is very difficult for people to act or come forward when they see unacceptable behaviour, unless they have help from their organizations. A “User-friendly Guide,” and the existence of a confidential organizational ombudsman may help people who are uncertain about what to do, or afraid of bad consequences for their speaking up.
Responsibility of journals
Journals are responsible for safeguarding the research record and hence have a critical role in dealing with suspected misconduct. This is recognised by the Committee on Publication Ethics (COPE) which has issued clear guidelines on the form (e.g. retraction) that concerns over the research record should take.
- The COPE guidelines state that journal editors should consider retracting a publication if they have clear evidence that the findings are unreliable, either as a result of misconduct (e.g. data fabrication) or honest error (e.g. miscalculation or experimental error). Retraction is also appropriate in cases of redundant publication, plagiarism and unethical research.
- Journal editors should consider issuing an expression of concern if they receive inconclusive evidence of research or publication misconduct by the authors, there is evidence that the findings are unreliable but the authors’ institution will not investigate the case, they believe that an investigation into alleged misconduct related to the publication either has not been, or would not be, fair and impartial or conclusive, or an investigation is underway but a judgement will not be available for a considerable time.
- Journal editors should consider issuing a correction if a small portion of an otherwise reliable publication proves to be misleading (especially because of honest error), or the author / contributor list is incorrect (i.e. a deserving author has been omitted or somebody who does not meet authorship criteria has been included).