Research misconduct is on the rise, certainly within hard science subjects, quite possibly elsewhere. Researchers around the world are inventing data, falsifying findings, and plagiarising the work of others. Part of this is due to the pressure on some researchers to publish their findings in academic journals. There is also career-related pressure on researchers to conduct accurate polls, produce statistically significant results, and get answers to questions, among other things. Some clients, managers, funders and publishers have a low tolerance for findings that chime with common sense or the familiar conclusion of ‘more research is needed’. They may expect researchers to produce interesting or novel findings that will direct action or support change.

Publishers are working to counteract misconduct in a variety of ways. Plagiarism detection software is now routinely used by most big publishers. Also, journal articles can be retracted (i.e. de-published) and this is on the increase, most commonly as a result of fraud. However, the effectiveness of retraction is questionable. The US organisation Retraction Watch has a ‘leaderboard’ of researchers with the most retracted papers, some of whom have had more papers retracted than you or I will ever write, which suggests that retraction of a paper – even for fraud – does not necessarily discredit a researcher or prevent them from working.

Some research misconduct can have devastating effects on people, organisations, and professions. People may lose their jobs, be stripped of prizes or honours, and be prosecuted in criminal courts. Organisations lose money, such as the cost of wasted research, disciplinary hearings, and recruitment to fill vacancies left by fraudulent researchers. And whole professions can suffer, as misconduct slows progress based on research. For example, in 2012 the Journal of Medical Ethics published a study showing that thousands of patients had been treated on the basis of research published in papers that were subsequently retracted. Retraction Watch shows that some papers receive hundreds of citations even after they have been retracted, which suggests that retraction may not be communicated effectively.

Yet even the potentially devastating consequences of misconduct are clearly not much of a deterrent – and in many cases may not occur at all. Let’s examine a case in more detail. Hwang Woo-Suk is a researcher from South Korea. In the early 2000s he was widely regarded as an eminent scientist. Then in 2006 he was found to have faked much of his research, and he admitted fraud. Hwang’s funding was withdrawn, criminal charges were laid against him, and in 2009 he received a suspended prison sentence. Yet he continued to work as a researcher (albeit in a different specialism) and to contribute to publications as a named author.

Closer to home, a survey of over 2,700 medical researchers published by the British Medical Journal in 2012 found that one in seven had ‘witnessed colleagues intentionally altering or fabricating data during their research or for the purposes of publication’. Given the pressures on researchers, perhaps this is not surprising – though it is deeply shocking.

The examples given in this article are from hard science rather than social research. Evidence of misconduct in social research is hard to find, so it would be tempting to conclude that it happens less and perhaps that social researchers are somehow more ethical and virtuous than other researchers. I feel very wary about making such assumptions. It is also possible that social research is less open about misconduct than other related disciplines, or that it’s easier to get away with misconduct in social research.

So what is the answer? Ethics books, seminars, conferences etc frequently exhort individual researchers to think and act ethically, but I’m not sure this provides sufficient safeguards. Should we watch each other, as well as ourselves? Maybe we should, at least up to a point. Working collaboratively can be a useful guard against unethical practice – but many researchers work alone or unsupervised. I don’t think formal ethical approval is much help here, either; it is certainly no safeguard against falsifying findings or plagiarism. Perhaps all we can do at present is to maintain awareness of the potential for, and dangers of, misconduct.

A version of this article was originally published in ‘Research Matters’, the quarterly newsletter for members of the UK and Ireland Social Research Association.