Data deposition as a measure to prevent and to detect scientific misconduct CERN workshop on...
-
Upload
maria-diana-hoover -
Category
Documents
-
view
213 -
download
0
Transcript of Data deposition as a measure to prevent and to detect scientific misconduct CERN workshop on...
Data deposition as a measure to prevent and to detect scientific misconduct
CERN workshop on Innovations in Scholarly Communication (OAI6)Geneva, June 17 – 19, 2009
Alexander LerchlSchool of Engineering and Science
Jacobs University [email protected]
The REFLEX Project
• Funded by the European Commission
• 2000 – 2004, ~ 3 M€ budget
• 12 research groups from 7 European countries
• Coordinator: Franz Adlkofer, Munich (Verum Foundation)
• 8 Publications from Austria (Medical University Vienna)
The First Case: Diem et al., 2005
Huge Effects on DNA Molecules andExtremely Low Standard Deviations
A Letter-to-the-Editor, Vijayalaxmi et al., 2006
...
The Reply: Rüdiger et al., 2006
Original data!
n = 2 for all means!
Last digit preference!
Out of 48 last digits:2: 14 times5: 2 timesp<0.001
Variations lower
than theoretical
limits!
24 average valueseach: n=2
Coefficients of variation always lower than 5%.
Accuracy of the method ~ 25%!
What did Mutation Research do?
• Nothing.
Onlin
e Feb
ruar
y 20
08
The Second Case: Schwarz et al., 2008
Again: Huge Effects and Negligible Standard Deviations
Coefficients of Variation
• Sham exposed: 3.9%
• Negative controls: 4.1%
• Exposed: 2.6%(at 25% variations of the SAR values!!)
• Positive controls: 1.2%
Please keep in mind: n=3!
Two Fortunate Coincidences
1) The discovery of Christian Wolf
2) The discovery of a student
The Discovery of Wolf: How to Unblind the Exposure System
The Discovery of the Student: Hidden Data
The student: „Just do a right-click on the graph, and open the Excel sheet. The original data are then visible“.
So I did ....
n = 12, and not n = 3 !
So, n = 12, and not n = 3!
Calculated “Raw“ Data
What did Int Arch Occup Health do?
• (Almost) Nothing.
Reactions of the Medical University Vienna
• In three press releases, the University called the data “fabricated“
• They informed the journals accordingly
Outlook
• None of the papers in question (altogether eight) has been retracted (so far).
• For example:
Outlook
• COPE (Committee on Publication Ethics) is currently considering a complaint against the journals [Mutation Research and IAEOH] to investigate whether they followed COPE's recommended procedures.
Conclusions
• These cases highlight the need for deposition of original data when a manuscript is submitted in order to make investigations possible if suspicions about the scientific integrity of submitted or published articles arise. This is of particular importance for papers dealing with topics of potential public concerns.
Thank you for Your Attention!
Questions?