More and more editors are faced with the issue of research misconduct—how to detect it and what to do if it is found.
Experts from the Office of Research Integrity (ORI) and the National Science Foundation’s (NSF’s) Office of Inspector General (OIG) offered advice and tips on dealing with “White coat crime, aka misconduct” when CSE gathered in Baltimore, MD, 1–3 May, for its annual meeting.
Susan Garfinkel provided an overview of their respective offices’ roles in battling misconduct.
Garfinkel said ORI falls under the US Department of Health and Human Services and the Office of the Assistant Secretary for Health. It is separate from the National Institutes of Health (NIH), even though a majority of the cases involve NIH research. The mission of ORI is to protect the integrity of Public Health Service–supported research. The office does not have oversight over research funded by foreign organizations or research that is funded privately, such as that done by pharmaceutical companies.
OIG’s goals, Kroll explained, are to provide leadership and coordination and to recommend policies to prevent and detect fraud, waste, and abuse. OIG’s jurisdiction covers NSF activities, programs, and operations. Basically, he continued, the organization exists to ensure integrity in NSF’s programs and operations. Kroll described research misconduct as a “growth industry,” noting OIG has found more misconduct in the last 5 years than in the previous 15.
Both OIG and ORI handle misconduct in a similar fashion. The two define misconduct as fabrication (making up data), falsification (manipulating or omitting data), or plagiarism (misappropriation of data without giving credit). Investigations done at these agencies stem from an initial allegation, which can come from other agencies, institutions, reviewers, students, or journal editors. Garfinkel stressed allegations must be credible and specific.
Following an allegation, an inquiry determines if evidence warrants an investigation. If so and the case falls under the jurisdiction of the agency, an investigation explores the allegation in detail. The investigation must be able to name the specific person responsible for the misconduct.
This process ends with a decision or case closure. It is considered serious if misconduct is found, Garfinkel said, and possible outcomes range from administrative actions to debarment. Kroll said the potential exists for civil or criminal remedies in some OIG cases. All case closeouts are made public on the agencies’ Web sites. The entire process can take some time, normally a year to 18 months, he added.
Garfinkel also touched on the increase in image manipulation in recent years and the Photoshop forensic tools ORI uses to detect it. She suggested session participants look for inconsistencies within an image, particularly in the backgrounds. The main approach in using Photoshop forensic tools, she explained, is to increase the visibility of “hidden details” that are not perceptible to the human eye. These tools include contrast enhancement, histogram equalization, embossing, gradient maps, and overlay of images. When conducting a panel overlay of two supposedly different images, there should always be some differences, and manipulation can be detected if they are “too similar to be different.”
Image manipulation occurs in many ways, she continued, including duplicating images within or between papers, within an image, or with alterations.
ORI and OIG can only investigate allegations within their jurisdictions. But other options are available when research misconduct falls outside of these agencies.
Elizabeth Wager discussed the global perspective, citing the UK Research Integrity Office as a resource for journals. The voluntary advisory body provides guidance and support regarding good research practices and how to address fraud and misconduct in research, although it does not have any statutory or investigative powers.
Ideally, Wager said, she would like to see a national, multidisciplinary oversight advisory body because “research misconduct doesn’t recognize funding boundaries.”
All presenters agreed journals are not in a position to investigate misconduct.
Garfinkel said when it comes to dealing with misconduct, “I can’t tell you what journals can do, but what questions they can ask.” She said journals should be asking if the raw data have been supplied, whether reviewers are getting the best-quality images, and whether production staff has the photo-editing programs necessary for catching image manipulation. Editors should ask whether a pattern is present and whether there are similar problems in other papers. They should question whether alterations arose in revisions and if the author attempted to negotiate. Garfinkel said journals should be suspicious if an author quickly offers to remove an image when questioned about it. Journals also should ask whether the author’s data or explanations are satisfactory or raise more questions.
The responsibility of investigating claims should fall on the institutions, Wager said. She advises institutions to consider misconduct investigations as a “badge of honor” instead of an embarrassment because they show recognition that misconduct exists and they are taking the correct steps to address it.
When finding the right person to contact within an institution, Wager suggested starting with the vice president of research, as that position often has oversight responsibility. She acknowledged dealing with misconduct can be “tricky,” but suggested sending a registered letter, rather than an e-mail, working with local board members, working jointly with other editors, and approaching professional or licensing bodies where available.
Finally, Wager said, persistence also is needed. She advised both institutions and journals to have clear policies in place to deal with misconduct.
Garfinkel suggested editors contact the institution and not the author, as in her experience, data tend to get “lost” once people are notified of an allegation. Garfinkel and Kroll both said if a journal refers a matter to ORI or OIG, the agencies are restricted by privacy acts as to what they can report back to a journal until a finding is made, which can take months. Institutions may be freer to communicate information with journals and may be able to take more immediate action. Kroll stressed that OIG would like to work with editors, but editors need to be aware of OIG’s limitations, including the fact its timeline is not the same as those of the editors.
The question of correcting the literature also arose. Some journals retract too quickly, before an investigation is complete, and many early retractions do not inform readers about misconduct issues, Garfinkel explained. Wager suggested an “editorial expression of concern” while an investigation is underway. This could prevent premature retraction yet still alerting readers that the research is in question. ORI and OIG publish their findings, but if misconduct is not officially found, the agencies can only suggest actions.