Presentation is loading. Please wait.

Presentation is loading. Please wait.

Safeguarding Science Against Research Misconduct

Similar presentations


Presentation on theme: "Safeguarding Science Against Research Misconduct"— Presentation transcript:

1 Safeguarding Science Against Research Misconduct
John E. Dahlberg, PhD Formerly, Deputy Director Office of Research Integrity (ORI) Office of the Assistant Secretary for Health Department of Health and Human Sciences USA Disclaimer: The views expressed here are mine and do not necessarily represent those of ORI or of the US government.

2 Terms Defined Research misconduct (RM)
Questionable research practices (QRP) Responsible conduct of research (RCR)

3 Outline First talk covers: Research misconduct
Questionable research practices Importance of scientific integrity education to promote Responsible Conduct in Research (RCR). Second talk covers: Mechanisms used in the US by institutions and ORI to investigate and handle allegation of research misconduct.

4 Increased Allegation of Research Misconduct
Availability of Adobe Photoshop Research becomes more complex and specialized  Requiring inter-lab, inter-disciplinary or even international collaborations  Increasing number of coauthors  Making it increasingly difficult if not impossible for all coauthors to bear equal responsibility for the entire paper. Availability of software to detect problematic images and plagiarism. PubPeer, Retraction Watch etc

5 Research Misconduct occurs in every major research institute in every country
ORI received visitors from: Australia, Canada, China, Croatia, Denmark, Germany, Japan, Korea, Singapore... Each country has many cases of misconduct. ORI co-hosted 4 international conferences on RM. “Don't go to a university that hasn't had at least one person fired for misconduct, because it means they are not looking for it properly.” (Elizabeth Wager, Managing Editor, Committee on Publication Ethics)

6 How Many Scientists Fabricate and Falsify Research?
1.9 % of researchers admitted to having falsified or fabricated research. 14.2 % had personal knowledge of a colleague who falsified or fabricated research data. Less than 50 % took action. Fanelli D (2009), PLoS ONE 4(5): e5738

7 ORI Statistics As of mid-2015 - greater than 6000 total allegations.
Annually - About 100 serious allegations Closes cases Makes about 15 findings of misconduct Many cases where misconduct occurred closed with ORI findings due to: Resource limitations Inability to prove which person responsible Minimal misconduct not justifying ORI findings Respondent disappearing (ORI is requiring to provide opportunity to respond to proposed findings)

8 ORI ”Findings” Remain Relatively Constant
ORI findings limited by resource and strict legal requirements to make a “finding”.

9 ORI STATISTICS: 1992 - June 2010 Total allegations 3622
Average allegations per year 198 Total cases closed 611 Total misconduct findings 220 Misconduct findings that involve clinical research 66 Findings leading to debarment 135 Retracted papers 159

10 Junior Investigators Committed More Scientific Misconduct
Rank Cases Mis-conduct % Department Chair 5 Professor 72 18 25 Associate Prof 29 40 Assistant Prof 66 27 41 Post Doc 96 50 52 Student 39 26 67 Technician 79 63 TOTAL 429 200 47 ORI Statistics

11 The pre-ORI era 1970s-late 1980s: Many (in)famous cases of Scientific fraud
William Summerlin at Sloan-Kettering, 1974 John Darsee at Harvard, 1981 Stephen Bruening at Pittsburgh, 1987

12 The Summerlin Case Dr. Summerlin at Memorial Sloan-Ketterling in New York City. Transplantation immunology In 1974 painted a white mouse with black ink Other issues were found later. “Painting the mouse” became famous as the same as “research fraud.”

13 The Darsee Case John Darsee, MD, worked at Emery University ( ) With Eugene Braunwald at Harvard ( ) Minor misconduct found at Harvard Harvard found no other problems (3 investigations) NIH (slowly) found much wrong with Darsee – and Harvard Much additional misconduct at Emory and earlier. 17 Retractions resulted.

14 The Stephen Breuning Case
Dr. Breuning - a psychologist at the University of Pittsburgh Studied the effect of drugs on mentally retarded children. Multiple investigations conducted by NIH and Pittsburgh. Dr. Breuning finally confessed Two years of probation Barred from received grant funds. The complainant (Dr. Sprague) was treated very badly.

15 The Creation of ORI A theme of early (pre-ORI) misconduct:
Inadequate investigations Poor treatment of complainants Led to Congressional pressure on NIH NIH created the Office Science Integrity (OSI) in 1989 OSI became ORI in 

16 ORI Has Two Divisions Division of Investigative Oversight (DIO):
Assist institutions handling allegations. Review institution’s reports and make separate findings Division of Education and Integrity: Responsible Conduct of Research Education and training Conferences Research on misconduct

17 The ORI definition of “Research Misconduct”
White House Office of Science Technology Policy Federal-wide definition of research misconduct Federal agencies to develop policies and procedures ORI issued a new regulation Code of Federal Register at 42 C.F.R. Part 93 “Research misconduct” = Fabrication, falsification, or plagiarism in proposing, performing, or reviewing research, or in reporting research results. Does not include honest error or differences of opinion.

18 Three Types of Misconduct
Fabrication: making up data or results and recording or reporting them. Falsification: manipulating research materials, equipment, or processes, or changing or omitting data or results such that the research is not accurately represented in the research record. Plagiarism: the appropriation of another person's ideas, processes, results, or words without giving appropriate credit. 43 CFR §93.103

19 Examples of Falsification
The two sets of dots outlined in red are identical. And each pair of dots outlined in yellow are internally identical

20 Falsified loading controls

21 Falsification/band duplication

22 Fabrication Fabrication can be more difficult to detect because there is no original data at all. Fabricated numbers can often be detected statistically using an ORI-developed software tool (DigiProbe)

23 Detection of Fabricated Numbers
These numbers were thought to be fabricated The software used tested only the two digits on the right of the number For example, only 9 and 8 from 5098 .

24 Results of Using DigiProbe
824 questioned digits (with no notebook records) were compared with 1021 digits from assays (with records). The two right hand digits are tested The probability that the questioned digits were not fabricated was < The probably that the other data were not fabricated was 0.78.

25 Plagiarism The use of other people’s language and/or ideas without giving due credit. Quoted language should be noted within quotation marks (”…..”) The source should be cited.

26 ORI handling of Plagiarism
ORI has licensed iThenticate for years to check questioned papers and grant applications for plagiarism ORI allows minor plagiarism cases to be handled by the institution. Although ORI considers data falsification and fabrication as more serious, and makes few findings of plagiarism, publishers of journals have a different view

27 iThenticate web site

28 Falsification, Fabrication and Plagiarism (FFP) vs. Research Misconduct
ORI must distinguish between detecting something false and stating that this represents research misconduct (a ‘Finding’). ORIs regulation states that it must be shown that the acts of falsification, fabrication or plagiarism have been done with some level of deliberate attempt to deceive.

29 Questionable Research Practices (QRPs)
Other practices, which do not fall under the definition of misconduct but may violate institutional rules or best scientific practice. Two papers discussing this topic are: “Scientists Behaving Badly,” a Commentary in Nature 435(9): , 2005, Brian C. Martinson, Melissa S. Anderson and Raymond de Vries Lex M. Bouter, et al., Research Integrity and Peer Review (2016) 1:17, DOI /s

30 Top 5 “Questionable Research Practice”
Rank Cause 1 Insufficient supervision of junior coworkers! 2 Inadequate handle or store data or (bio)materials 3 Use of published ideas or phrases of others without referencing 4 Keep inadequate notes of the research process 5 Ignore basic principles of quality assurance 2016 Router et al.

31 Research misconduct vs. Questionable research practices
Research misconduct occurs in every country Much RM is not reported. Questionable research practices (QRPs) are more common QRPs are not research misconduct. How to improve science and minimize RM and QRPs? Teaching Responsible Conduct in Research (RCR)

32 Nine Core Areas for Responsible Conduct in Research (RCR)
Human Subjects Mentor/ Trainee Responsibilities Research misconduct Animal Welfare Data Acquisition, Management, Sharing, & Ownership Collaborative Science Publication Practices & Responsible Authorship Conflict of Interest and Commitment Peer Review 3 1 2 4 8 5 7 6 9 Historically, education and the responsible conduct of research has centered on 9 Core Areas of RCR Instruction (Figure 2). The quest for research excellence series of booklets will explore each of these areas in greater detail and additional areas of RCR instruction (Figure 3).

33 Consequences of Research Misconduct
The Respondent’s career may be ruined. Institutional reputation at risk. Public research funds wasted. Investigations can take years and are expensive. Attempts by the field to reproduce false research are wasteful. Trust in in science is compromised.

34 Conclusions: How to Prevent Research Misconduct: Advice for the PI
Enhanced mentorship Demand proper record keeping and retention. Review raw data, not just ‘finished’ figures. Be suspicious

35 RCR Resources Thank you!
The ORI web site contains a vast array of RCR resources A list of links to ORI’s RCR resources Interactive DVDs (“The Lab” available with Chinese subtitles) RCR training modules provided by multiple institutions Thank you!


Download ppt "Safeguarding Science Against Research Misconduct"

Similar presentations


Ads by Google