The scientific enterprise places high value on honesty and openness. These and other values serve it well. Nevertheless, a variety of challenges to the integrity of scientific research have been cause for much recent soul-searching.
There is growing concern, for example, that substantial percentages of published results in some fields, biomedicine and psychology in particular, are not reproducible. In the meantime, new deleterious practices have arisen, such as predatory new journals that do little or no peer review or quality control and yet charge authors big fees. And retractions of scientific papers are rising. These are just three of the problems addressed in Fostering Integrity in Research, a 284-page report by the Committee on Responsible Science of the National Academies of Sciences, Engineering, and Medicine. The study was sponsored by a number of science-oriented agencies.
“The research enterprise is not broken, but it faces significant challenges in creating the conditions needed to foster and sustain the highest standards of integrity,” said Robert Nerem (Georgia Institute of Technology), chair of the committee. He called on everyone involved in science to “take deliberate steps to strengthen the self-correcting mechanisms that are part of research…”
The report endorses the definition of scientific misconduct in the 1992 Academies report Responsible Science: “fabrication, falsification, or plagiarism in proposing, performing, or reporting research.” However, many other practices, such as misleading use of statistics, that fall short of fabrication and failure to retain and share research data have been categorized as “questionable” practices. The report says these should now be considered “detrimental” research practices and strongly shunned.
Journals and research institutions both have a responsibility to curb irresponsible or abusive actions. The report says they should go beyond simple compliance with federal regulations to instead monitor research conduct and sustain the highest standards. In addition, good faith whistleblowers who raise concerns about the integrity of research must be protected and their concerns addressed fairly, thoroughly, and in a timely manner according to the report.
The report says standards for transparency in many fields do not adequately support reproducibility and the ability to build on previous work. All information needed for a person knowledgeable about the field to reproduce the reported results must be made available at the time of publication or shortly after, the report says. This has been a long-standing problem in some fields (see adjacent story, “A Scientist Pushes Psychology Journals toward Open Data”). The new report says funding agencies and other sponsors should allocate funds to enable the long-term storage, archiving, and access to datasets and codes necessary to replicate findings.
The fact that a growing percentage of recently published research turns out to be not reproducible is one of the report’s main concerns. The rising irreproducibility of published results is a complex phenomenon. While a certain level of irreproducibility due to unknown variables or errors is a normal part of research, the report says data falsification and detrimental research practices such as inappropriate use of statistics or after-the-fact fitting of hypotheses to previously collected data also play a role.
Researchers should routinely disclose all statistical tests carried out, including negative findings, the report says. It points to studies showing that scientific publications are biased against presenting negative results and that publication of negative results is on the decline. But there are powerful reasons for publishing negative results, the report emphasizes. Routine reporting of negative results helps avoid unproductive duplication of research and makes research spending more productive. Also, publishing negative results can prompt questioning of established paradigms, leading ultimately to groundbreaking new discoveries. Everyone involved should support and encourage this level of transparency, the report maintains.
On the problem of a rising number of retractions of journal articles due to research misconduct, the report cautions that this may not be due to a rise in the rate of misconduct; a more vigilant scrutiny by the scientific community could contribute. But there is no doubt that for a variety of reasons, “research misconduct and detrimental research practices are becoming more visible and attracting more attention” (p. 65).
While most of the report focuses on the institutional duties of research organizations, funding agencies, and scientific journals, it also challenges individual scientists to uphold the highest level of conduct possible. “Each individual retains deeply personal obligations and duties to aspire to the highest levels of rigor in his or her work,” the study says (p. 65). These include “understanding about cognitive biases and errors” to build in precautions against favoring existing ideas. Personal biases, interests of funders, and the reward systems surrounding the pursuit of science also must be considered.
Another issue addressed is authorship. The report calls on scientific societies and journals to develop clear authorship standards based on the principle that those who have made a significant intellectual contribution are authors. All authors should approve the final manuscript. Gift or honorary authorship, coercive authorship, or ghost authorship should all be universally condemned, says the report.
The report calls for a new independent nonprofit organization, a Research Integrity Advisory Board, to be set up. It would work with all parties to promote high ethical standards and minimize research misconduct and detrimental research practices.
The report may or may not help restore public trust in science, now under attack from many quarters. But by reemphasizing the need for new institutional practices for self-examination and quality-control, it shows that scientific leaders are aware of the new challenges to research integrity and are striving to make changes to minimize them.
The new report can be downloaded or read online at https://www.nap.edu/catalog/21896/fostering-integrity-in-research.