UNITED STATES

Academies’ efforts to close the gender gap pay off – Study

Efforts by both the American Academy of Arts and Sciences (AAAS) and the National Academy of Sciences (NAS) to address historic bias against female researchers in the United States have largely been successful, says a new study published in the Proceedings of the National Academy of Sciences of the United States of America (PNAS) in January 2023.

Prior to the 1980s, for six sample disciplines, including economics, mathematics and cellular biology, the percent of women scientists inducted into the American academies was below 10%; the one sample discipline where female inductees was above 10% was the NAS’s anthropology category in which almost 20% of new academy members were women.

In 2021, by contrast, in economics, which has historically had very low levels of female inductees, more than 50% of the AAAS’s inductees were female, reports the study, “Gender gaps at the academies”.

“There’s a long-standing concern that women’s contributions in science are under-appreciated and that may be one of the reasons why young women are a little bit less likely to pursue STEM [science, technology, engineering and mathematics] fields. And there’s also the feeling that they can feel discouraged by the lack of female role models,” said Professor David Card, Class of 1950 Economics Professor and director of the Center for Labor Economics at the University of California, Berkeley.

Card was co-winner of the Nobel Prize for Economics in 2021 for his work on the effects of the minimum wage, immigration and education on the labour market. “Gender gaps at the academies” is Card’s inaugural PNAS article after being elected to the NAS in 2021.

“The motivation for our project was to see to what extent it was true that females are under-represented in the academies and whether this has changed over the past 10 or 15 years when the academies started to make efforts to ensure gender diversity of the candidates and make sure they weren’t overlooking distinguished people who should be admitted on the basis of their work.”

For his part, Professor Stefano DellaVigna, one of Card’s three co-authors, who also teaches in the department of economics at the University of California, Berkeley, told University World News it was “extremely important” to understand inequities in the academic process as well as anywhere else in society.

“Are women or different ethnicities being treated differently? How do we know that women are under-represented in some particular disciplines? How do we know if it is due to discrimination or, maybe, the pipeline is missing?” he asked.

“If we look, for example, at why there are relatively few female PhD students, it’s a little hard to tell what it’s due to. It’s difficult to know whether these students [who were not admitted] should have been admitted.

“So that’s why we look at honours – because 20 or 30 years into someone’s career, you have a really good measure of the accomplishments of that person: the number of publications in the most important journals in the field and citations to that work. The American Academy and National Academy are high-level honours. They don’t go to people that are very junior, by definition.”

Maths, economics and psychology

Assembling and then analysing their database of 85,367 authors in psychology, 38,994 in economics and 91,424 in mathematics spanning six decades was a massive undertaking that began with determining the highest impact journals in each field.

Using Google Scholar profile pages, Card, DellaVigna and their two co-authors, Economics Professor Patricia Funk from Università della Svizzera italiana in Lugano, Switzerland, and Ikerbasque Research Professor Nagore Iriberri from the University of the Basque Country in Bilbao, Spain, identified the most cited journals in psychology, economics and maths, the three disciplines they examined statistically.

Among the journals judged to have the most impact are the Journal of Personality and Social Psychology, Child Development, American Economic Review, Journal of Economic Theory, Annals of Mathematics and Communications on Pure and Applied Mathematics.

Their second step involved downloading the titles and authors’ names for every article published in these journals between 1930 and 2019; they did not include notes, letters to the editors and articles of less than three pages.

Using this, they created “a synthetic annual CV for researchers who were listed as an author of one or more of the selected articles”, they write.

Next, using a list of names prepared by the United States Social Security Administration, they determined which authors were female and which were male. Since some names, such as Andrea, which is female except among Italians, can apply to both genders and because there are a relatively small number of females in some categories, to ensure that they did not miss them, they employed a number of research assistants to go through the names that remained “ungendered”.

Maths had the highest number of authors of unknown gender because of the tradition of maths journals identifying authors by their first initial and family name instead of by their given name and family name.

In the final step, the research team matched the data set of the annual CVs to the list of newly selected members of the NAS and AAAS in each of the three test disciplines.

In a given year, there could be 15,000 researchers in economics, for example. Most would not have anywhere near the publications that could get them elected. But, DellaVigna said, there are a good number of people with 10 or 20 publications that look like they could be elected to the National Academy of Sciences.

And, since the researchers knew (with a high degree of accuracy) who was male and who was female, they plotted if the election rates for professors with similar publication and citation records differed by gender.

(A proof of the robustness of their statistical analysis is that in 2022, using the same statistical procedures, Card et al were able to predict with more than 50% accuracy the economists who were elected that year as fellows of the Econometric Society.)

Evidence of early discrimination

One of their key findings is that, starting in the 1970s, when their dataset is more robust, there is clear evidence of discrimination against female professors: neither academy elected a female mathematician until the mid-1980s. Between 1960 and 1979, every one of 103 mathematicians elected to the AAAS and the 75 inducted into the NAS – an average of nine persons per year – was male.

According to DellaVigna, even controlling for the lower level of female maths professors, there are fewer articles written by females and, thus, “it definitely looks as though women had a harder time publishing”. While this could mean that when they did publish, that publication was likely worth more, he said. However, women did not have the number of publications necessary for election.

Card and his co-authors’ study does not include qualitative data, such as interviews to help determine the politics of the academic eco-system of the periods they study. Still, DellaVigna said, their findings “definitely replicate what we hear from researchers who were around back then [pre-1980]: that women mathematicians were treated very poorly and were not taken seriously”.

In economics the story is similar. Between 1960 and 1979 there were no females elected to the NAS and only 1.87% of the 107 inductees of the AAAS were female. Over the next two decades the percentages of women inducted were 5.26% (of the 38 inducted) and 3.68% (of 138), respectively.

Echoing what DellaVigna said about maths departments, Card told me: “I’m in my late 60s … I met quite a few of the first women to be tenured [in economics].”

In the first two decades of this century, again, paralleling mathematics, the percentage of women economists being selected as fellows of the academies has grown substantially. However, since the numbers of younger female economists in the pipeline has not grown substantially, Card et al characterise the jump to about 15% for each academy reflecting “a change in the probability of selection conditional on qualification”.

Efforts to include under-represented groups

When I ask Card what surprised him in the study, he said it was the improvement in the number of female psychology researchers elected to the academies.

Between 1960 and 1980, in the discipline of psychology, 5% of the 40 inductees into the NAS were female and 7.14% of the 56 inductees into the AAAS were female. Between 1980 and 1999, the percentages were 15.63% (of 32 inductees) and 25.71% (of 70 inductees), respectively.

Card et al’s analysis of the publication record of highly qualified candidates revealed that between 1960 and 1989 “it appears that the female share of newly selected academy members was roughly in line with the female share of highly qualified candidates”.

Between 1990 and 2005, they write, the number of female psychology scientists elected to the academies rose from about 20% to over 40%. In the two decades ending in 2019, of the 106 researchers elected to the AAAS, 39.62% (or 41) were females; 27 of the 55 scientists inducted into the NAS were female.

Part of the reason for this increase is a rise in the number of female researchers with 15+ major publications. In previous decades the percent of female researchers in what’s called “the pipeline” hovered between 5% and 15%. By 2020, this figure had risen to 30%.

This increase is not, however, enough to explain the rise in the number of female scientists elected to the academies. Rather, Card and DellaVigna explained, the academies have worked to include previously under-represented demographic groups.

“Thus, in this later period and especially in the more recent decade, the female share among newly elected members lies substantially above the female share among highly qualified researchers, pointing to a [statistical] preference for female members.”

In our interview, Card further explained the increase in the number of female scientists in maths, economics and psychology inducted into the academies by noting that the team’s data shows that “there is a little bit of a tendency to push them into the academy at slightly younger ages”.

Conscious that his words could be taken to mean that the academies are watering down standards, before referencing the ages of the female inductees, he said: “When you look at the candidates who are being nominated and voted in as fellows, they are extremely distinguished. I think the main thing that’s happening is a kind of acceleration of the process so that they are elected somewhat younger.”

International changes

“Gender gaps at the academies” focuses on American scientists only, so its findings are not necessarily generalisable to other national academies.

However, Card told me that in the study “Gender differences in peer recognition by economists,” published in October 2022 in Econometrica, the database of which was international, the results were broadly similar.

Referring to a number of countries, he said: “They’ve done a very good job of expanding representation and not just of females – but also of under-represented countries, particularly in Asia, Africa and South America. The situation is much different than it was even 15 years ago, and the academies deserve recognition for their success in increasing the diversity of fellows.”