Women wait around ten years in America for a diagnosis of endometriosis, a painful but historically little studied gynaecological condition. Some medical devices, such as pulse oximeters, used to detect drops in oxygen levels, do not work as well for people with dark skin. Might such delays and biases reflect skews in academic research? A recent paper published in PNAS, an academic journal, noted imbalances in various fields. The study estimates that if researchers over the past 40 years had been reflective of the American population, there would have been 29% more articles on public health, 26% more on gender-based violence, 25% more on gynaecology and 18% more on mental health.

The authors, Diego Kozlowski, Vincent Larivière, Cassidy Sugimoto and Thema Monroe-White, collected the names of more than 1.5m researchers in America who published papers between 2008 and 2019. Using data on how frequently various first names and surnames occurred in different genders and races, Mr Kozlowski and his colleagues then applied an algorithm to predict, for each researcher, the likelihood that they were male or female and of a given race. Combining these probabilities with the papers’ topic of study, the authors estimated how many researchers of each race and gender were publishing in a given field.

The authors found that as well as an overall lack of representation of women, black and Latino people in academia, researchers of different races and genders tended to cluster within certain research topics. While white men were the most evenly spread across academia, women, black and Latino researchers tended to study topics that received fewer citations. Women are disproportionately likely to publish in fields related to health and psychology; they remain less likely to publish in maths, physics and engineering. (Asian women buck this trend, concentrating in chemistry and biomedicine.)

The paper did not investigate the extent to which these disparities reflect choice or discrimination. But women and minorities working in prominent fields tended to have fewer citations than their white, male peers. Citations are often used as a measure of a researcher's influence, which is important for securing funding. White applicants are more successful than minorities in grant applications and men tend to receive more funding than women. This may drive women and minorities into less prestigious fields. Research could look very different if the researchers did, too. ■

For a look behind the scenes of our data journalism, sign up to Off the Charts, our weekly newsletter.