Are Bad Data Management Practices The Root of Research Misconduct?

ReportonResearchCompliance.com  October 2006 Heard the one about the researcher who wrote his equations related to a new strain of organism on paper towels?  Seems the data were accidentally sacrificed for the greater good of a clean lab counter.

This true story is a good (perhaps extreme) example of what is formally called “bad data management.”

Bad data management can scuttle promising research and cause trouble in other ways.  In fact, most scientific misconduct allegations stem from poor data management practices, according to Chris Pascal, director of the Office of Research Integrity (ORI).

For many years, ORI has wanted to communicate more directly with researchers.  The paper towel story came to light at the first such conference, “New Capabilities, Emerging Issues, and Responsible Conduct in Data Management,” held Sept. 28-29, 2006, in Baltimore. 

Cosponsored with the University of Maryland, the conference drew participants from throughout the United States – and even from Denmark, according to Stephanie Bird, a consultant and co-editor of the journal Science and Engineering Ethics who organized the conference for ORI.  In addition to sharing information about good data practices, ORI made the conference interactive to elicit information about what standards for data management exist – or don’t – in universities today.

“The purpose of the conference was to actually talk to researchers about what their standards and expectations are with reference to data management,” Bird explained.  “The goal was really to talk about an aspect of research for which there are variable or no regulations.”

(1)     Don’t review the raw data prior to publication; accept summary data or prepared tables or graphs instead

(2)    On a project where expected results have not been achieved over several months, demand significant results by Friday to meet a publication or grant deadline.

(3)    Hire a new post-doc who comes highly recommended, but leave him or her without guidance or supervision.

(4)    Tell your staff to do the right thing, but do the convenient thing when it is expedient.

(5)    Publish results of a team research project, but leave out one or two members of the team who made substantive contributions.

(6)    Provide no guidance or standards for keeping laboratory data.

(7)    Tell your lab members to ask questions, but don’t make yourself available because you are too busy.

(8)    Have a large lab of junior scientists and provide little guidance or supervision.

(9)    Drop data points in order to clean up your graph or table without a clear rationale.

(10)            Tell your lab tech what results you expect from the experiment and that you need the results right away.

In addition to hearing from other speakers, conference attendees broke in small group to strategize about the proper way to handle two common scenarios:  a dispute over authorship and patent rights, and suspicions that emerge when a researcher cannot duplicate the results of an investigator who is no longer at that institution.

To reach a broader audience and promote discussion among researches and administrators, Bird said that papers from the conference presentations and discussion well be published in the journal she co-edits.

The first meeting for researchers was considered a success, and others are being planned.  ORI originally planned a conference for 150 people.  Because of the size of the conference venue, ORI limited registration to 170 and turned many others away, Bird said.

Larry Rhoades, ORI director of the division of education and integrity, Pascal, and Bird are planning the next one, which will occur sometime in 2007.