When Jennifer Jacquet first visited Carl T. Bergstrom's evolutionary-biology lab at the University of Washington last year, she was surrounded by men. Men staring at data on the 27-inch Mac Pro computer screen that takes center stage in the lab. Men talking about mathematical proofs, about a South Park episode on evolution, about their latest mountain-climbing adventures.
Use an exclusive Chronicle tool to search gender differences in nearly 1,800 fields using historic data analyzed by the University of Washington's Eigenfactor Project.
Jennifer Jacquet, now a clinical assistant professor in environmental sciences at NYU, made the suggestion that kicked off the gender project, which she then worked on. The results showed that "things are getting better for women in academia," she says.
Carl Bergstrom (left), a theoretical biologist at the U. of Washington, and Jevin West, a postdoctoral researcher there, analyzed whether the authors of two million scholarly articles were male or female.
"The lab is like visiting a fraternity," says Ms. Jacquet, who completed her postdoc at the University of British Columbia before starting this year as a clinical assistant professor in environmental studies at New York University.
Perhaps being the only woman in the lab prompted Ms. Jacquet's answer when Mr. Bergstrom, a professor of theoretical and evolutionary biology, asked her what should be done with a remarkable new trove of data. Mr. Bergstrom and Jevin D. West, a postdoc, had just gotten their hands on nearly eight million scholarly articles collected by JSTOR, a digital archiving service. They went all the way back to Isaac Newton's time.
The biologists at Washington, who specialize in using mathematical models and computer simulations to see how academic ideas flow through networks of scholars, wondered how they could use the data.
For Ms. Jacquet, the answer was clear: What did the articles and their authors show about gender differences in publishing? Were women and men equal in this fundamental coin of the academic realm, a currency that buys tenure, promotions, and career success?
To Ms. Jacquet's surprise, Mr. West and Mr. Bergstrom took her idea and ran with it. The result is the largest analysis ever done of academic articles by gender, reaching across hundreds of years and hundreds of fields. "This has never been done on this scale before," says Mr. West.
Although the percentage of female authors is still less than women's overall representation within the full-time faculty ranks, the researchers found that the proportion has increased as more women have entered the professoriate. They also show that women cluster into certain subfields and are somewhat underrepresented in the prestigious position of first author. In the biological sciences, women are even more underrepresented as last author. The last name on a scientific article is typically that of the senior scholar, who is not necessarily responsible for doing most of the research or writing but who directs the lab where the experiment was based.
"What we've done is assemble this huge collection of data across many of the major scientific, social-science, and humanities fields, providing a new lens for looking at how gender plays out in scholarly authorship," says Mr. West. "But this also provides a platform for further research to be done. We don't want this to stop here."
Publication Can Mean a Job
Scholarly publishing, more than anything else, is the measuring stick of professors' research productivity. In the humanities, it's usually the monograph. But in the hard sciences and in many social sciences, it's journal articles.
To be hired on the tenure track in those fields by a top research university, young scholars increasingly must have publications on their CV's by the time they finish their doctoral degrees. And once they are hired, more publications in leading journals typically are required to be promoted at every step along the way to full professor.
"When I went to get my first job, I had a sole-authored paper in a very good journal," says Shelley J. Correll, who started her academic career as an assistant professor of sociology at the University of Wisconsin at Madison in 2001. "I think that article is what undoubtedly got me that first job. And it's even more common now for students to have that."
Ms. Correll is now a professor of sociology at Stanford University and director of its Michelle R. Clayman Institute for Research on Gender. Her name will appear fourth on an article that Mr. Bergstrom's lab hopes to publish on the findings, behind Mr. West (who is first), Ms. Jacquet, and Molly M. King, one of Ms. Correll's graduate students. Fifth is Theodore C. Bergstrom, an economist at the University of California at Santa Barbara, and last is his son, the University of Washington's Mr. Bergstrom.
Author order is very important. "If I were to give people a vita of two people who had the exact same number of publications and one person was first author on a lot of papers and the other had publications in the same journals but was second through fourth author, I guarantee you people will prefer first," says Ms. Correll.
So negotiating author order becomes crucial. But women may not be as confident and have as much experience as men with those negotiations. "If I'm writing with a man, he may be more likely to insist he be first," Ms. Correll says. "When women negotiate in general, they are less likely to be successful. People don't consider their requests as legitimate."
That made names and gender the first order of business for Mr. Bergstrom's lab.
First they created an algorithm to label the millions of JSTOR papers by field and subfield. Then the trick was to figure out whether an author was male or female. The lab consulted data on birth names collected by the Social Security Administration. If a name was used at least 95 percent of the time for a female, they coded it female, and the same for a male. If use of the name was more ambiguous, they threw the paper out.
Of the eight million articles the group started with, it ended up analyzing two million—written by 2.7 million scholars—whose author composition was similar to the whole. Roughly half were published between 1665 and 1989, and the other half between 1990 and 2010. Included in the database are papers in the hard sciences, the social sciences, law, history, philosophy, and education. Missing from the JSTOR data are articles in engineering, English, foreign languages, and physics.
The data show that over the entire 345 years, 22 percent of all authors were female. (Even though few papers in the JSTOR archive originated in the first 100 years, the researchers still felt that examining the entire data set was worthwhile.) The data also show that women were slightly less likely than that to be first author: About 19 percent of first authors in the study were female. Women were more likely to appear as third, fourth, or fifth authors.
According to the data in just the most recent time period, it is clear that the proportion of female authors over all is rising. From 1990 to 2010, the percentage of female authors went up to 27 percent. In 2010 alone, the last year for which full figures are available, the proportion had inched up to 30 percent. "The results show us what a lot of people have been saying and many of my female colleagues have been feeling," says Ms. Jacquet. "Things are getting better for women in academia."
Women still are not publishing, though, in the same proportion as they are present in academe as professors. The same year that the share of female authors in the study reached 30 percent, women made up 42 percent of all full-time professors in academe and about 34 percent of all those at the most senior levels of associate and full professor, according to the American Association of University Professors.
As the proportion of female authors over all has grown, the biologists' study found, so has the percentage of women as first authors. In fact, by 2010 about the same proportion of women were first authors as were authors in general—about 30 percent.
But those gains have not been mirrored in the last-author position, which is of particular importance in the biological sciences. According to the data, in 2010 only about 23 percent of last authors over all were female. In molecular and cell biology, women represented almost 30 percent of authorships from 1990 to 2010, but only 16.5 percent of last authors. And over that same time period in ecology and evolution, women represented nearly 23 percent of authors but only 18.5 percent of last authors.
"The gap between women as first authors and women as last authors is actually growing, which suggests that women in scientific fields are allowed to have ideas and do most of the work on a paper, but do not yet have the big grants and labs full of students and postdocs that would establish them in the prestigious last-author position," says Ms. Jacquet.
A peek beneath the surface, going from field down to subfield, also reveals hidden disparities. For example, the database shows that over the final 20 years covered by the study, about 23 percent of authors writing in ecology and evolution have been female. But in some subfields of those disciplines, the proportion of women authors was even smaller, including just 19.5 percent in herpetology and 16.6 percent in paleontology.
The same phenomenon exists in economics: Between 1990 and 2010, 13.7 percent of authors in the database were female. In some subfields, though, the proportion of female authors was even smaller. In macroeconomics, for instance, women represented just under 10 percent of authors. In a subfield labeled "household decision-making," however, the proportion of female authors shot way up, to 30 percent.
What's notable, says Mr. Bergstrom, is the way some of those differences mirror gender-role stereotypes. For instance, in sociology, where the gender breakdown was more even—women represented 41.4 percent of authors since 1990, men 58.6 percent—things look very different in the subfield of criminology where men made up 74 percent of all authors. On the other hand, women were much more heavily represented in subfields including "sex roles," where females accounted for 56 percent of the authors, and in the study of aging parents, where women constituted 62 percent.
"Sociology represents a domain where we think of things being more even because of the overall gender ratios," says Mr. Bergstrom. "But once you get in, for whatever reason you see we stratify ourselves. And we do it in ways that are loosely consistent with our common sense, the stereotypes about who does what."
Why Not 50-50?
Women's progress in academe has long been a hot topic, not least the debate over why women publish less than men do. Female professors are more likely to emphasize quality over quantity, some scholars argue, turning out fewer but meatier pieces than do their male colleagues, who are more apt to increase their productivity by publishing their work in more-frequent chunks.
In addition, studies show that women spend less time on research and more time on teaching and committee work. And it is often research and publishing, which require sustained attention, that suffer when women devote time to caring for young children.
As more women earn Ph.D.'s and take faculty jobs, though, and as the gap between the number of women and men in academe narrows, scholars have begun thinking about whether anything can or should be done about gender-based differences that remain in publishing, hiring, promotion, and pay. Do those differences result from choices women make, scholars wonder, or from discrimination?
Where professors come down on that question influences how they interpret the data on scholarly publishing from Mr. Bergstrom's lab. Cheryl Geisler, dean of the Faculty of Communication, Art, and Technology at Simon Fraser University, in British Columbia, says the fact that women do not account for at least half of the professors in most fields, and half of those writing scholarly articles, doesn't necessarily mean that women face discrimination. But "if it's not 50-50," she says, "we should at least ask, Why not?" Authorship data showing that some subfields are even more male or female than the average is "like the canary in the coal mine" and deserves further study, she says.
Some academics argue that gender clustering in subfields can skew the results of scholarship. "Just as when you had primarily Europeans writing history, you had a certain version, the same thing happens when women begin to write about crime and punishment," says Anita Levy, associate secretary of the AAUP. "There may be a way of thinking about it, or women who may see fit to look at different factors that may not have occurred to a scholar who is male."
Ann Mari May, a professor of economics at the University of Nebraska at Lincoln, has done research showing that gender influences the way economists view social policy. She sent surveys to 400 members of the American Economic Association who hold doctorates in economics from American universities. Of the 143 who responded, she found that the women among them were 24 percentage points more likely than the men to believe that the size of the U.S. government is either about right, too small, or much too small, and 21 points likelier to reject the idea that the United States suffers from excessive government regulations.
John J. Siegfried, a professor emeritus of economics at Vanderbilt University who has worked with the economic association, says there is not much that academe can do about gender imbalances, short of forcing women into subfields that may not interest them. "What's the solution?" he asks. "Should I tell women starting a Ph.D. that they should only study finance or econometrics?"
Wendy M. Williams is a professor of human development at Cornell University who studies the role of women in science, including scholarly publishing. She says the Bergstrom data on gender and authorship don't necessarily show discrimination in academic publishing, even though women cluster in some subfields and publish at lower rates than either their male counterparts or their presence in academe in general.
"The international literature show that when women submit work, there is no bias in it being accepted, but the likelihood of women submitting work may be lower," she says. "If a woman is interested in a field, but she has to devote time to three kids, she may not be submitting as often. I don't see that as discrimination."
Just last month, though, a new study released by Yale University suggested that gender bias remains a factor in academe. It found that science professors at six major research universities were likely to rate male job candidates as more qualified than female candidates to be hired as laboratory managers, even though the study assigned the hypothetical male and female applicants identical qualifications.
Ms. Jacquet says she was surprised to find that her own experience as a female scientist was not reflected by women in the Washington study of the JSTOR archives. At the time she visited Mr. Bergstrom's lab last year, she had published 10 scholarly articles and was the first author on all of them. Her fellow postdoc Mr. West, she noticed, had published a few more in total but was the second author on five. Although she knew that being first author was important, Ms. Jacquet also wondered whether it was because she is female that she had had to take the initiative in publishing all of her own scholarly articles, while Mr. West had been asked to collaborate on several peer-reviewed articles.
But the data, she says, show that female professors in the study actually were more likely to be second through fourth authors than first. It knocked down her theory that male scientists had failed to ask her to collaborate on academic articles because she is a woman. Since she first visited Mr. Bergstrom's lab, in fact, she has published three academic articles on which she is not the lead author. The article on gender and authorship will be her fourth.
"For me," she says, "this really showed the beauty of science, that you can have this personal experience that isn't reflected in big data."
Correction (10/22/2012, 3:36 p.m.): This article originally misstated Ms. Jacquet's authorship status on 10 scholarly articles she had published as of last year. She was the first author on all 10, not the only author. The article has been updated to reflect this correction.