Last month, in a rare pronouncement by the US scientific establishment on a thorny subject, the Institute of Medicine (IOM) released a report on integrity in scientific research.

The IOM notes that fully fledged cases of scientific misconduct are rare. But it sensibly calls for research institutions to take a more active role in creating an environment where misconduct will not occur.

Every scientist would agree that good science requires solid experimental design, truthful and thorough reporting of results, honest peer review, good care of living subjects, and fairness to one's colleagues and students. The IOM report looks at what can be done to encourage researchers to adhere to such nostrums throughout their careers.

It concludes that government mandates are unlikely to imbue ethics. Nor are classes in which students are herded into a lecture hall to fulfil a requirement on a checklist. Advising and mentoring are crucial but, as the report states, the quality of mentors varies immensely. So it suggests that “both the call for change and its implementation must come from research institutions”.

But having accepted that, what are universities to do? The IOM suggests that they teach research ethics creatively, as an integral part of core course content. It recommends that professional bodies evaluate institutional integrity as part of the accreditation process. However, no one knows how institutional characteristics influence research integrity, and the report calls for more research on this linkage.

This assessment comes hard on the heels of two recent cases of alleged misconduct, at the University of California, Berkeley, and at Bell Labs in Murray Hill, New Jersey. Institutions would be wise to take its premise seriously. Whenever a researcher commits major fraud, he or she has probably been getting away with smaller lapses for years — fudging a control here, deleting a messy data point there.

It is in the interests of every university and laboratory to help students think through the long-term consequences of what may at first appear to be minor violations of integrity. Ultimately, such early consideration will be good for institutions, good for the careers of young researchers, and good for science itself.