Paper faces pattern of all colors. Black history month celebration of diversity.

Authors are being asked to apply the highest standards of rigour.Credit: Getty

It is regrettable but true that researchers have used and abused science to justify racist beliefs and practices. As previous editorials have acknowledged, Nature has played its part in perpetuating racism — and has now pledged to play its part in tackling it, together with colleagues in the research community.

As part of this pledge, Nature and the Nature Portfolio journals are updating our advice to authors on reporting research that involves race, ethnicity and other socially constructed characteristics (see Specifically, we’re asking that authors exercise care and consideration so that the highest standards of rigour are applied where these attributes are found to be an explanation for an outcome or conclusion. This is part of our ongoing updates to guidance asking authors to describe how demographic characteristics, including sex and gender, are considered in the design of studies — and, more broadly, to consider the research’s potential to cause harm.

There are many important reasons to study race or ethnicity. People of colour are affected by discrimination in education, at work and elsewhere. These situations need to be — and are being — studied, so that problems can be better understood and solutions can be found.

But there are instances in which a focus on race or ethnicity as an explanation for an outcome can be inaccurate, and has the potential to be harmful. For example, studies of human behaviour sometimes attribute differences to race or ethnicity when they could involve other variables, such as socio-economic status or occupation. Inaccurate inferences that race is the decisive variable run the risk of entrenching stereotypical attitudes — to the detriment of the communities involved.

So, what are we asking authors to do, if their research describes people according to race, ethnicity or other socially constructed categories? Essentially, three things. First, specify the categories used and explain why such classification is needed. Second, explain the methods used to describe people in this way — for example, did study participants self-report, or did the information come from a census, social media or administrative data? Third, we would like authors to describe how they controlled for confounding variables, such as socio-economic status. These requests will be added to a paper’s reporting checklist so it is a part of the usual editorial and publishing workflow.

Explaining methods of classification is important because race and ethnicity are not fixed categories, and there are many proxy measures for them. For example, one increasingly common method is to train machine-learning algorithms to assign people’s race according to their names. Such categorizations can be inaccurate because they reflect naming conventions, rather than having anything to do directly with a person’s heritage.

The research enterprise is on a path towards stopping discrimination and ensuring equity. In updating our advice, we are joining other journals on this journey. Overall, we all want authors to think harder and more carefully about these issues. The advice we’re announcing this week, and our existing measures, are intended to uphold the highest possible standards of rigour and accuracy in research. They are also meant to keep research from inadvertently perpetuating harm, and to avoid creating more negative experiences for people for whom racism is a daily lived reality.