How much misconduct should there be in research?

That's a strange question. Of course there should be no misconduct in research.

The problem is that it's difficult to know exactly how much misconduct – data fabrication, plagiarism and so on – there actually is. Instead, we have to infer this from the number of allegations that are made, the number of these that are upheld following investigation and so on. In other words, our evidence is indirect.

And that leads us to the question of how much misconduct there should be. If no allegations are made, or no allegations are upheld that might be for two reasons. One is simply that there is no misconduct. 
I'm not sure that's plausible. Another is that misconduct isn't reported, or properly investigated. That's more plausible…

According to the 2019 UK Concordat to Support Research Integrity (a revision is expected this year), UK institutions are committed to: 'upholding the highest standards of rigour and integrity in all aspects of research; ensuring that research is conducted according to appropriate ethical, legal and professional frameworks, obligations and standard; supporting a research environment that is underpinned by a culture of integrity and based on good governance, best practice, and support for the development of researchers; using transparent, timely, robust and fair processes to deal with allegations of research misconduct should they arise; and working together to strengthen the integrity of research and to review progress regularly and openly'.

Institutions are also required to prepare an annual report, which should include a breakdown of the number of allegations of misconduct received. Unfortunately, a UK Committee on Research Integrity (UK CORI) report found that for 13 per cent of institutions surveyed a report could not be identified online, and was not available on request, for any of the three academic years investigated. They conclude, perhaps optimistically, that 'As we do not have evidence as to whether these statements have been written, we were not able to draw conclusions around alignment with Concordat expectations (i.e., the statements may exist but have not been shared online or with us)'.

The fact that reports could not be found for so many institutions isn't exactly encouraging. But what happens when we look inside the reports that do exist? The UK CORI report tells us that around 60 per cent of institutions reported at least one misconduct allegation in any given year within the period investigated (2019-2022), and around 50 per cent reported at least one misconduct investigation. Think about that. In any given year 40 per cent of institutions report having not received one single misconduct allegation.

So, either misconduct is pretty rare. Or something else is going on…

In 2012, Leslie John and colleagues surveyed over 2,000 psychologists about their involvement in questionable research practices. They triangulated evidence from three different estimation methods and concluded that the prevalence of falsifying data at 9 per cent. In other words, the prevalence of questionable research practices, including those that constitute clear misconduct such as falsifying data, is surprisingly high.

Other evidence reinforces this conclusion. Elizabeth Bik – a 'data sleuth' who received the 2024 Einstein Foundation Individual Award for raising awareness of questionable research practices – has estimated that 4 per cent of articles in her field have problematic images. And the Dutch National Survey on Research Integrity found that 8 per cent of Dutch scientists had committed research fraud by fabricating data.

That would suggest misconduct is not that rare. At least not as rare as institutional annual reports would have us believe. So what else might be going on? It's difficult to know, but it's likely to be a number of factors.

One is that institutions may feel that misconduct allegations would reflect badly on them, and try to minimise the numbers they report – for example by deciding that some allegations are vexatious and therefore excluding them from their numbers.

Another is the processes may be opaque or difficult to navigate. An interesting exercise is to try to find out from your institutional website exactly how to submit a misconduct allegation. Is it clear and straightforward? Are the protections in place similarly clear?

And another, related factor might be the extent to which people within an institution feel that they can trust the process, however clear and straightforward it might be. Do they fear possible retribution? 
Will whistleblowing end badly for the whistleblower?

If we are to truly know what the prevalence of research misconduct is, we need to ensure there is a culture of psychological safety that makes anyone involved in the research process to raise concerns. 
We need to make that easy to do (without encouraging trivial or malicious accusations, of course). And we need institutions to be transparent and open about their processes, the number of allegations they receive, and how these are investigated.

And we should also think about ways of making investigations themselves feel less threatening. 
An investigation of potential research misconduct can be clearly as damaging to individual researchers (as well to the university itself). Normalising the investigation of research misconduct in less accusatory terms is also important. A report by the UK Research Integrity Office (UKRIO) concluded exactly this – that what is required is a 'A … system that promotes transparency, destigmatises allegations of research misconduct, and normalises early raising of concerns'.

This is easier said than done of course. It is currently in everyone's (narrow, short-term) interests to keep things quiet. It's easier for institutions, funders and journals to maintain the impression of very low levels of misconduct. Change will require courage, particularly on the part of these organisations, to genuinely tackle misconduct, and recognise that ultimately this will improve the quality of research, and therefore the speed with which this research advances knowledge and benefits society.

For me, the findings of the UK CORI report – both the number of institutions for which no annual report could be found, and the high prevalence of nil reports of allegations – suggest we have some way to go. If something seems too good to be true, it probably is… Sadly, I doubt that research misconduct is vanishingly rare.

Marcus Munafò is Deputy Vice Chancellor and Provost at the University of Bath, and Associate Editor for Research.

What research/metascience ideas would you like to see discussed in these pages? Send your suggestions for topics and people to the editor at jon.sutton@bps.org.uk.