Ethics: Detecting misconduct

Nature (2006) | doi:10.1038/nature04996

Does a digital workflow make it easier to detect ethical breeches in peer review?

The Internet has changed everything. You can be sitting at your desk in Birmingham, Alabama, while having a conversation in real time with a colleague in Birmingham, United Kingdom, exchanging not only words and ideas, but also photographs, data sets and manuscripts. The Internet has also changed the way science is done, particularly when it comes to publication. Manuscripts are now submitted, reviewed and authors notified electronically. But although the efficiency and speed of the peer-review process has increased, a set of attendant issues has arisen.

Specifically, it is now easier to detect breaches of ethical behaviour than ever before. As evidence, the number of reported ethical problems involving publications of the American Physiological Society (in 14 separate journals) rocketed from an average of less than one a year before 1999 to more than 50 a year in 2004, when all of the society's publications became available online1.

Whether this is coincidental or causal is open to debate, but I contend it is the latter. When evaluating a manuscript, a reviewer no longer has to trek several blocks to the library to scour the printed journals in search of a paragraph or a figure that seemed familiar. All that has to be done now is to type a few keywords into an appropriate search engine and, hey presto, all the relevant articles will appear on your desktop. Or, for the more meticulous, antiplagiarism software is available for free download.

Other, more draconian misconduct-detection measures are aimed at identifying image manipulation. These are currently being considered and even implemented by some journals. A number of forensic tools that scrutinize scientific images are available from the US Department of Health and Human Services' Office of Research Integrity (see 'Forensic software traces tweaks to images' and 'Should journals police scientific fraud?').

Image processing

It is all too easy for authors to manipulate images for publication. For example, digital image-processing programs make it a simple matter to prettify ugly gels. Unwanted background, smudges and 'non-specific' bands can be easily removed from the final figure. I have always thought that showing only a single band of interest in a figure such as a western blot or immunoprecipitation experiment is a somewhat equivocal practice, although I admit I have done it (often at the urgings of an editor or reviewer to present data in as concise and beautiful a manner as possible).

But if we all agree that the foundation of good science is accurate, reliable and reproducible data, then we must be prepared to accept images that are less than perfect. If images are manipulated to enhance what we aim to demonstrate, even if our intentions are good, we chip away at the integrity of the scientific enterprise and erase the trust that the public places in our work. Both the Journal of Cell Biology2 and Nature3 should be applauded for aiming "to end the fetish of the perfect image".

The computer and Internet have forced us to define new parameters to guide the way in which data are acquired, stored, analysed and presented. Certainly there are legitimate reasons why figures are manipulated (for example, altering the order of lanes in a gel to match the order of lanes in a comparable gel). But maybe it is time to simply take the high road: if you want the gel to look different, do the experiment again. That way, any criticism about data manipulation is deflated.

I do not think it prudent for journals to become active detectives, seeking out potential scientific misconduct in every submitted manuscript. That would impose an unnecessary confrontational relationship on authors and publishers, even before the process of peer review began. Moreover, the considerable time and resources that journals would need to muster in order to police potential misconduct would be prohibitive.

Journals may be responsible for ensuring the integrity of the scientific record, but trust in our scientific colleagues and ourselves has been, and should continue to be, the underpinnings of our discipline. If we, as individuals, do not police our own actions and actively instruct our students in proper behaviour, you can bet someone else will.

REFERENCES

1. Benos, D. J. et al. Adv. Physiol. Educ. 29, 59�74 (2005).

2. Rossner, M. & Yamada, K. M. J. Cell Biol. 166, 11�15 (2004).

3. Nature 439, 891�892 (2006).

RELATED LINKS:

Office of Research Integrity

Dale Benos is an endowed professor and chair of the University of Alabama at Birmingham's Department of Physiology and Biophysics. He is a former editor-in-chief of the American Journal of Physiology-Cell Physiology and is or has served on the editorial boards of nine biomedical journals. He is currently a member of the Publications Committee of the Federation of American Societies for Experimental Biology.

Visit our peer-to-peer blog to read and post comments about this article.

Extra navigation

.
ADVERTISEMENT