Saturday, January 28, 2012

The emperor is marching around buck naked/BMJ

The validity and quality of research underpin the entire research enterprise worldwide.  However, a number of studies have shown that many researchers take "shortcuts" and that perhaps 1-3% of research is grossly false, fitting into the category of research misconduct.

Research misconduct has been defined in US federal law as fabrication, plagiarism and/or falsification.  Identified cases have been few, often limited to easy-to-identify falsified figures in published papers.

However, these may be the tip of the iceberg.  “Ginny Barbour, a senior editor with the PLoS group of journals, said one third of authors could not find the original data to back up figures in scientific papers when these were questioned.”  OUCH!  Have 1/3 of authors fabricated or falsified data?

Certainly, the incentives to do so are great:  career advancement in a highly competitive environment and obtaining grants that enable you to perform more research.  While the disincentive -- being publicly identified and having one's career destroyed -- occurs so infrequently the risk is negligible.

As John Noble notes below, the measures currently in place to guarantee reliability of research (peer review and professional ethics) fail miserably.  Additional standards are desperately needed.
Godlee’s report of widespread research misconduct is disturbing and, I believe, generalises to the US.1 2 It’s not just the fact that it is happening, but that it reflects a culture within which new researchers are socialised. Previous research indicates the motivators for dishonesty include a high pressure achievement oriented environment, where “if everybody else is doing it, it must be OK.”3
The problem also underscores how important are reanalysis and replication of reported research, a recent topic on the US Institutional Review Board Forum (www.irbforum.org/forum/). But, get this, “Ginny Barbour, a senior editor with the PLoS group of journals, said one third of authors could not find the original data to back up figures in scientific papers when these were questioned.” I wonder how many journal editors seek to discover and reject such authors. I wonder how meticulous the US Food and Drug Administration is in policing the input it receives in support of marketing approval for new drugs and medical devices.
These behaviours are beyond the reach of surveillance by institutional review boards and research ethics boards, and they indicate the need for a new end product quality control system.4 The simplifying assumptions about the efficacy of peer review and professional ethics and responsibility fail miserably when the emperor is discovered marching around buck naked.

No comments: