White men have more freedom over research choice – Study
The study, ‘Intersectional Inequalities in Science’, published in Proceedings of the National Academy of Sciences last month, examined the names of the authors of more than 5.4 million articles published in the United States archived in the Web of Science (WOS) database.
“There’s a strong statistical affinity between the fact that you’re part of a minority group or a woman and the fact that you study specific topics. And those topics suffer from a citation bias (ie, are cited less often)” and, therefore, have less ‘impact’,” said study co-author Vincent Larivière, Canada Research Chair on the Transformations of Scholarly Communication at Université de Montréal.
By contrast, he said: “White men have the privilege of choice,” summed up in the belief, “I can work on whatever I want.”
A long way from a meritocracy
According to his co-author, Thema Monroe-White, based at the Campbell School of Business, Berry College, Mount Berry, Georgia, the findings show that American academia is far from being the meritocracy many claim it is.
“Were science a meritocracy,” she told University World News, “there would not be a relationship between race and gender, research topic and impact. For example, in the social sciences, black women authors were more highly associated with the topics of ‘racial discrimination’ and ‘gender-based violence’ than other first authors [ie, the lead author of an article].
“In the health fields, we see this same relationship on the topic of ‘racial disparities’. These findings imply that, for minoritised scholars, our racialised and gendered identities shape what we study.”
Minoritised scholars and topics are also the least cited, she said, suggesting that “advances in scientific knowledge are shaped by socially constructed [(ie, race and gender)], non-meritocratic factors that reinforce marginalisation of topics [and, therefore, authors] in science”.
To arrive at their findings, Larivière, Monroe-White and their two co-authors, PhD student Diego Kozlowski (faculty of science, technology and medicine at the University of Luxembourg) and Cassidy R Sugimoto (professor of public policy at the Georgia Institute of Technology in Atlanta), disaggregated the data in the WOS using an algorithm that extracted keywords such as children + schools in titles and abstracts to create what they call a ‘topic space’. Another algorithm uses US census data to infer the race and gender of each article’s lead author.
Larivière admits that last names are not a perfect indicator of race. Nevertheless, since the US census requires residents to indicate how they self-identify out of six racial groups, such as ‘White, Black, Asian, Latinx, American Indian and Alaska Native’, and two races [biracial], as well as the respondents’ family names, the census provides a probabilistic means to determine the race of the more than 1.6 million lead authors of the research papers.
“The majority of people named Rodriguez, for example, self-described in the census as Latinx” said Larivière.
Because disaggregating names for black Americans is more difficult due, in part, to the legacy of slavery in the US, Monroe-White said it was more difficult to assign a probabilistic value for race using name-based algorithmic approaches. “Ultimately, this may lead to an inflated estimate of the number of black authors in a given topic space,” she said.
The bibliometric analysis has, nonetheless, been welcomed by some researchers.
Dr Donna K Ginther, who directs the Institute for Policy and Social Research at the University of Kansas, said it provided the hard numbers that show how minoritised individuals’ career paths as researchers and the topics they study within specific fields are overdetermined by their life experience as minorities.
“The results are not surprising to me, having worked in this space for quite a while, but I think it is important to actually do the heavy lifting to show that this [the racialisation of research articles] is true in the data,” said Ginther.
The study shows that, while the percent of Latinx scholars rose from 3.5% to 6% between 2005 and 2018 and black scholars rose from 5% to 6%, both groups are under-represented as lead authors, especially in physics, mathematics and engineering. Female scholars are also under-represented in these fields, while they are over-represented among the articles on health, psychology and the arts.
By contrast, Larivière, Monroe-White et al found that, “Asian women follow a different pattern, with under-representation in arts, humanities and social sciences and an over-representation in biomedical research, chemistry and clinical medicine”.
The huge database allowed the researchers to drill down and show that, even within a topic space, scholars are balkanised.
For example, Asian authors who wrote on health, disproportionately studied China, proteins, cells and the economics of health. Black authors focused on the disparities of race and sexually transmitted diseases. More specifically, black women scholars tended to focus more on African American communities as a whole, while black men scholars published more about gay men.
Latinx authors publish more on topics that mention the Latinx population, racial disparities (a topic shared with black authors) and English-Spanish, according to the article.
Across all disciplines, the study found that minoritised groups are over-represented in lowly cited topics and under-represented in highly cited topics, meaning that the minoritised scholars’ work had less impact.
In health, for example, articles written by white males were cited more frequently than any other group, and white female professors saw their articles cited more often than their minority female counterparts.
However, white female professors’ articles were cited less frequently than Asian, Latinx and black professors, which indicates what might be thought of as a discount female scholars pay for their gender.
The fact that minoritised female professors are cited less frequently than their male minority colleagues, Larivière, Monroe-White, et al, write, “provides evidence of the intersectional between- and within-topic disadvantage for minoritised groups”.
White men – most ubiquitous across topics
When you look at places where there are fewer women (and minorities), said Larivière, they are “typically the disciplines that are most abstract and prestigious”, as measured by both grant monies and popular culture. Accordingly, white men are over-represented in physics, engineering, mathematics and chemistry. Yet, Larivière, Monroe-White, et al found that, even so, white men are “the most ubiquitous across topics”.
There are two reasons for this. First, white men make up the largest group of lead authors in the WOS database. Second, and more importantly, white males have the privilege of choosing high prestige topics or those topics which simply interest them.
According to Larivière, the privilege given by American society to white males is such that they are conceptually free to consider the entire realm of academic questions.
“If you are a white man, you are structurally free to go and study the topic that you want – and that will probably be the thing that everyone finds important and interesting. Other social groups do not have that level of privilege. So, they don’t necessarily work on the things that are the most important.”
Central to understanding the difference between white males and minority or female scholars, Larivière, Monroe-White and their co-authors argue, is the late French sociologist Pierre Bourdieu’s concept of ‘scientific capital’.
Like capital as traditionally understood, ‘scientific capital’ is not distributed evenly throughout the population. The structural racism of American society and, especially, its secondary schools, means that the foundation of scientific capital – being raised in a family which emphasises education and has the financial wherewithal to ensure enrolment in a good high school, which, for example, offers advanced placement maths and science classes – is not part of the lives of minoritised students.
Once in college or university, a student’s scientific capital builds on itself, as she or he is seen by her or his professors as being a likely member of the scientific field. Since these ‘gatekeepers’, as Monroe-White calls them, are disproportionately white males, what we might call a social version of the biological law ‘ontology recapitulates phylogeny’ (the individual of a species follows the same phases of evolution as the species itself, and thus recapitulates the species’ structure) kicks in as the professoriate recapitulates its ethnic and gender make-up.
Further, while considering itself committed to the neutrality of the scientific method, simply by teaching what constitutes interesting scientific questions the existing professoriate reproduces its (white) social assumptions.
Interestingly, Larivière said their data showed a double effect in the history of women scientists.
“The more women join a profession, the more that profession will be depreciated. So, it’s a double relationship, in the sense that you are a minority, so you’re unable to go into certain topics. But if you were there, it leads to the topic losing some scientific capital.”
Costs of stratification
Pointing to the fact that companies with more racially diverse workforces had higher profits and more diverse groups outperform homogeneous ones in cognitive tasks, Larivière said: “What we are criticising here is the notion that there is a free market of ideas and that everyone is free to work on the topics that they want.
“What we’re actually showing is that the thing you’re going to study is influenced by who you are.” A more diverse set of scientists “will probably be of higher quality and maybe have a higher social relevance”.
According to Brandeis Marshall, founder of DataedX, a company that works to counteract the automated oppression effects of data that is collected without culturally-responsive analysis, and who is also the author of numerous articles on the structural racism in much of the digital world, bringing someone’s identity into the research “means you have the ability to see the research and outcomes and observations in a much broader spectrum that is inclusive of all people.
“So, for instance, when it comes to the area of computing, it very much centres on whiteness and male whiteness. This means that the algorithms are designed in order to promote or amplify white people.”
For example, she told me, the first versions of facial recognition software were built to recognise white men with glasses. The algorithm could not recognise people with long hair, freckles or had a disfigurement, for example.
As with most conventional scientific articles, ‘Intersectional Inequalities in Science’ is replete with statistics, technical terms and charts and graphs. The article also has links to two websites, one presenting the data in depth and another that explains the methodology. Unlike most conventional articles, however, the discussion contains a counterfactual scenario.
According to Ginther, the counterfactual scenario is methodologically necessary because Larivière and Monroe-White et al’s work is an observational study that does not have access to a randomised control group. What you can do, she said, is “make a simple set of assumptions about what research would come out of demographic change”.
“I think that’s an important thought experiment about what scientists who study this topic think might happen in a different scenario.”
After assuming similar productivity and controlling for the ages of the lead author of the scientific papers, Larivière, Monroe-White and their co-authors estimated that, if the distribution of authors matched the 2010 US Census, in the absence of racial and gender inequalities, there would have been 29% more articles in public health, 26% more on gender-based violence, 25% more in gynaecology and in gerontology, 20% more on immigrants and minorities, and 18% more on mental health.
“While this counterfactual scenario is coarse, it highlights the fact that a different body of knowledge would be produced in the absence of inequalities and that this body would more closely reflect the spectrum of topics relevant across society. The diversification of the scientific workforce is necessary to create a scientific system whose results benefit all of society,” said Larivière.
What, then, do the authors of ‘Intersectional Inequalities in Science’ propose? Like Ginther and Marshall, Larivière, Monroe-White et al call, not only for more funding but, equally importantly, a redistribution of funding. “Funding agencies can take immediate action to allocate increased funding in areas that have been historically under-represented,” they write.
Monroe-White also argues for the need to “incentivise inclusion”. It is not enough for the STEM fields to emphasise recruitment and retention. To achieve both, they must find ways to make retention, the experience of being in the field, relevant to the racial or gender identity of the new entrants, she said.
“We know that there are topics where minoritised scholars are publishing; picking those up and amplifying them can lead to increased recruitment and retention in fields that have particularly struggled with this.”
The authors also call for the promotion of a “diversity of resources and initiatives that provide marginalised populations [access] into high-prestige networks”.
Expanding on this point, Marshall said that well-known papers are all too often cited even when doing so is not necessary. “Just because a paper is popular one year, doesn’t necessarily mean that the paper needs to be cited by everyone when another paper with a similar point or different perspective is available.”
Continually citing the same works means that “when people are in the classroom and they are discussing papers, the type of paper they’re discussing needs to be from a diverse set of known contributors but also perspectives,” she said.
Unravelling what Marshall calls the “citation paradox” will go some way towards broadening the funding, and, hence, fostering diversity. For, it means that when a minoritised individual is trying to get funding, he or she “will not be fighting those individuals and fighting those popular papers, which often sideline their [the minoritised professor’s] grant application.”