Acts of scientific fraud, such as fabricating or manipulating data, appear to be surprisingly common but are underreported to university officials, says a report published today in the journal Nature. And the institutions may have investigated them far too seldom, the report’s authors write.
The Nature report draws on the largest and most-systematic survey to date about research misconduct as defined by the federal government—namely, fabrication, falsification, and plagiarism. The Office of Research Integrity, a federal agency that oversees misconduct cases, sponsored the study. It was carried out with the help of the Gallup Organization, which collected responses from 2,212 federally financed scientists about apparent misconduct that they had directly witnessed among colleagues.
Extrapolating from the survey findings, the authors offered a “conservative” estimate of 2,325 possible instances of illegal research misconduct nationally per year. Of those, only 58 percent, or roughly 1,350 incidents, were reported to institutional officials. The authors call this small percentage “alarming.”
Based on the volume of observed misconduct, the authors argue that the number investigated by universities is too low. Federal rules give institutions that receive federal grants the lead responsibility for probing allegations against their researchers, but universities and other institutions have reported an average of only 24 investigations annually to the Office of Research Integrity. The office has the power to disbar scientists from participating in federally financed studies.
“Our study calls into question the effectiveness of self-regulation,” the authors write in a peer-reviewed commentary in Nature. “We hope it will lead individuals and institutions to evaluate their commitment to research integrity.”
The authors are Sandra L. Titus, an official in the research-integrity office, Lawrence J. Rhoades, the emeritus director of its education division, and James A. Wells, director of research policy at the University of Wisconsin at Madison. Mr. Wells previously worked for Gallup, where he directed the survey on research misconduct.
Their estimated incidence of misconduct is in line with those in a handful of previous studies. (The authors reported the incidence rate as at least 1.5 observed cases per 100 researchers annually.)
Questions About Methodology
But some observers criticized those previous estimates as seemingly too high and the studies’ methodologies as flawed. So the research-integrity office designed the survey and its study to respond to the criticism. For example, members of the authors’ research team evaluated whether the apparent misconduct described by the scientists surveyed appeared to meet the federal definition of research misconduct.
The leader of a previous major study on the topic called the latest one “sound and rigorous.” Brian C. Martinson, a senior research investigator at HealthPartners Research Foundation, a nonprofit organization in Minneapolis, led a 2005 study, also published in Nature, that found an even broader incidence of ethically questionable research practices, not just the federally proscribed kind (The Chronicle, June 9, 2005).
At least one university official still had questions about the new study in Nature. Robert R. Rich, the medical-school dean at the University of Alabama at Birmingham, said that, although he had not seen the study, the reported incident rate seemed high.
The authors cited several possible reasons for the disconnect between the number of offenses they estimated and the number of investigations by universities. For example, institutions may have determined that some of the allegations lacked merit. Other colleges may have resolved the complaints through an inquiry, a precursor to a formal investigation. Universities are not required to report the results of inquiries or how many they conduct to the Office of Research Integrity.
However, the authors say these do not “sufficiently explain the discrepancy.” They point a finger at institutions, saying that many may discourage reporting of research misconduct to avoid the resulting expense of investigation as well as the internal controversy and public embarrassment. The authors point out that the research-integrity study found that less than a third of institutional policies on misconduct explicitly obligate faculty members to report scientific misconduct. Another study found that 43 percent of whistle-blowers surveyed said their institutions pressured them to drop such allegations.
Mr. Martinson added that the study raises important questions about the ways that universities handle misconduct after an initial complaint.