Fraud: who is responsible?

Editor's Note -- The Office of Research Integrity (ORI) recently penalized two endocrinology researchers for fraud, including University of Indiana grad student Emily Horvath and Boris Cheskis, a former Wyeth scientist. Cheskis and his coauthors retracted two publications (including one that was highly cited), and two publications that Horvath co-authored are in the process of being retracted, including one published by Endocrinology. Jeffrey Blaustein, Editor-in-Chief of Endocrinology, has written the following opinion about scientific fraud, but notes that "the allegations that led to action by the ORI did not come from the editors or the Editor-in-Chief of Endocrinology, nor by the reviewers or readers of the journal."

Jeffrey D. Blaustein

Who is responsible for the fraudulent data making its way into publication? The Office of Research Integrity (ORI) recently reported two misconduct cases in which scientists committed fraud in research grant applications, and in one case, in papers published in peer-reviewed journals and even in a doctoral thesis. In the comments posted in response to a recent article on this website about one of these cases, blame was placed on the trainee, the principal investigator, the reviewers and the editors. Some commenters suggested that the student is a scapegoat.

While I cannot speculate on the roots of the specific cases, I suppose that in particular cases, any or all of these individuals could be responsible by reason of negligence. In some situations, the head of the laboratory may not be sufficiently critical of a trainee's primary data. In others, the reviewers or editors may not be sufficiently critical of data in a figure or a table. But let's look at two examples more closely.

First, with an electrophoresis gel, an errant trainee or collaborator can make a spot appear or disappear in either of two ways. After the gel is run, the scientist can easily manipulate the digital image. Fortunately, there are often tell-tale signs that are obvious at gross examination (e.g., splicing), and detection software is available that makes manipulated components of an image stick out like a sore thumb. Alternatively, a clever (but unprincipled) scientist can make a spot appear or disappear by altering what substances are applied to the gel. If done cleverly enough, it is possible that this would be undetectable.
Certainly, the head of the laboratory and collaborators cannot be expected to hover over their trainees or colleagues at all times.

With many other forms of data collection, if the collaborator or trainee were to simply change a value or two in a spreadsheet, there may be no sign visible to the head of the laboratory, collaborators, the journal reviewers or the editors. Fortunately, because the final step in the scientific method is replication, detection in both of these cases should come at the time of attempted replication. Unfortunately, this may be after the paper is published, a great deal of funds and effort spent on experiments and grants submitted based on the fraudulent data, and the paper extensively cited.

Fortunately, for another type of fraud, detection has gotten easier; plagiarism often gets uncovered. Just as professors can now identify plagiarism in student term papers by the electronic tool of Googling a sentence or phrase that does not seem to be the student's work, I and others have found plagiarized sections of published and unpublished journal articles by doing the same.

Colleagues have asked me, as Editor-in-Chief of a major research journal, if fraud is more prevalent or more likely to be discovered than in the past. I cannot answer the first question. The digitalization of science has made some types of fraud easier to perpetrate, but only marginally. Certainly it does not take much skill to alter digital images, but it also has never taken much skill to enter fabricated numbers in the results of an assay or other data sheet either. Likewise, retyping a paragraph from another author's work takes only marginally more time than cutting and pasting from a digital file.

Scientists who commit fraud, like all cheaters and thieves, believe they will get away with it, and unfortunately some do in the short term. As Editor-in-Chief of Endocrinology, I receive allegations of fraud from reviewers and readers, and occasionally from lead authors or other co-authors. The journal is a conduit for these allegations, but it is not an investigative body. Our first step is to discuss the issue with the authors. This often leads to the next step, which is to ask the institution's office responsible for compliance to investigate the allegation. Fortunately, many institutions take these allegations seriously, investigate and then report their findings to the journal, so that we may take appropriate action. If fraud is determined, this will involve retraction of the manuscript. If US Public Health Service funds were involved, as they were in the two recent cases, the institution may bring the charges to the attention of the ORI of the National Institutes of Health, which may then investigate and possibly take punitive action.

How might the scientific community prevent research fraud? Sadly, even if everybody performs their supervisory and oversight roles to the best of their abilities, some scientists will still engage in fraudulent behavior. The reasons that scientists engage in misconduct are as varied as the reasons that people steal, and that is a subject for others to write about.

This is not to say, however, that we should accept fraud. Scientists at all levels and in all capacities, from co-workers and PI's to reviewers and editors, must remain vigilant. Most importantly, when data are suspect, they must be investigated by the appropriate body and not swept under the rug. With the cooperation of everybody involved, the system works, and science is eventually self-correcting. This process is unfortunately sometimes necessarily slow. I have not figured out how those impatient individuals among us are to deal with this, but as a scientist, I have confidence that scientific truth will eventually prevail.

Whatever the causes for the fraudulent data, and however they were detected, the fact that they were discovered suggests that the system works, albeit sometimes too slowly and after much wasted time and energy of others.

The views expressed are those of the author and do not reflect those of the Endocrine Society.

Jeffrey D. Blaustein, PhD, is a Professor in the Center for Neuroendocrine Studies, Neuroscience and Behavior Program and Psychology Department of the University of Massachusetts, Amherst. He currently serves as Editor-in-Chief of Endocrinology and is President-elect of the Society for Behavioral Neuroendocrinology. His research group studies the cellular mechanisms underlying the interactions of the environment and stress on behavioral response to ovarian hormones.

Related stories:

Retracted: highly cited paper
[26th April 2010]

PhD student admits misconduct
[15th April 2010]

Plagiarism retracts review
[1st April 2010]