Start a conversation about research integrity and many researchers will assume you’re talking about misconduct. Too often, they are wrong. Research misconduct encompasses fraud, fabrication and plagiarism. It is essential to deal with such dishonesty thoroughly and fairly, but it’s patching up a tear after the damage is done. Research integrity includes such investigations, but it is much more. It is about creating systems that boost the quality, relevance and reliability of all research.
The distinction is clear at the 6th World Conference on Research Integrity, being held this week in Hong Kong. Yes, there are sessions on misconduct — but there are many more on improving science overall. The biggest impact on research integrity is achieved through sustained improvements in day-to-day research practices — better record-keeping, vetting experimental designs, techniques to reduce bias, rewards for rigorous work, and incentives for sharing data, code and protocols — rather than narrow efforts to find and punish a few bad actors. (Both are important, of course, and sometimes the same policies can address both problems.)The conflation of integrity and misconduct is problematic because it stops researchers from talking about ways to improve their work. Experts in quality assurance or stringent protocols sometimes avoid using words such as ‘rigour’ and ‘integrity’ for fear of alienating their colleagues by suggesting that their work lacks these qualities. One programme set up to encourage practices such as randomization and blinding in animal experiments was advised to change its name from a “research improvement project” to a “research optimization project”. This is ridiculous. No one should be arrogant enough to think that their research cannot be improved. Conducting research with integrity, honesty and accuracy is something to which every scientist should proudly aspire. And research integrity is not a virtue that is either present or absent: it is a capacity that can be enhanced. It requires ongoing training for both early-career researchers and more senior faculty members, as well as the reworking of perverse incentives, and rewarding researchers for important but unglamorous contributions. Every conversation or course in research integrity is an opportunity to grow. Expectations, skills and standards must be constantly updated, especially as new fields such as metaresearch reveal more about the practices that strengthen science.History bears this out. The idea of informed consent is less than 100 years old. In the Nature research journals, checks for blinding and randomization are just over six. Expectations that data and code will be fully open are becoming mainstream. These are all examples of improving research integrity. We should acknowledge them, and seek more.