From the Chronicle of Higher Ed., MARCH 06, 2016

By Paul Voosen

W

hen it comes to science, women and men remain unequal. And while stories about overt harassment dominate the news, a host of researchers are teasing out the subtle reasons for why inequalities exist.

Cassidy R. Sugimoto is one of them. An associate professor of informatics at Indiana University at Bloomington, Ms. Sugimoto is an expert at prying stories from the data hidden in the authorship pattern of studies. She has now discovered a way to peer back into the structures of labs themselves. And she’s been surprised by what she’s seen.

As a forthcoming paper by Ms. Sugimoto and three others will show, women are disproportionately performing the experimental work involved in producing science — the pipetting, the centrifuging, the sequencing. Men, meanwhile, are more likely to be credited for analyzing data, conceiving experiments, contributing resources, or writing the study. Women, it seems, are providing the labor of science.

Ms. Sugimoto’s paper, which will appear in Academic Medicine, represents a new front in the examination of gender in science. The team pulled its data — 85,000 papers from 2008 to 2013 — from seven journals of the Public Library of Science, a publisher that requires studies to break down their authors’ contributions into five categories. It’s a new tool, and one that will help refine the study of why gender gaps persist in science.

While the study finds sex a significant factor in contributor role even when controlling for other important variables, like time since first publication, it does not explain why this disparity exists. Are women choosing to stay in these roles, or are the structures of science forcing it upon them? It’s impossible to say, Ms. Sugimoto said.
"There’s certainly still a gender problem in science," she added. "And this paper begins to provide a piece of the problem."

Across the sciences there’s been a flourishing of research looking more closely, with more data and increased experimental rigor, at the position of women in science. These studies, many of them done by early-career scientists, are teasing out a more nuanced story than individual headlines might blare, one that highlights unconscious bias and stereotype threat, and that also documents gains women have made when it comes to early-career hiring and measures of productivity.

These studies tend to come in two flavors: There’s exploratory work with huge data sets, like Ms. Sugimoto’s study, to discover actual trends in science. Such work, however, can rarely discover the causes behind divides. Experimental work, meanwhile, can discover instances of unconscious bias, but such results are difficult to connect to large trends. Given human nature, scientists, journalists, and readers tend to fill this explanatory gap with their own conclusions.

"Everybody is ready to document the gender gap," said Michele Pezzoni, an economist and an assistant professor at the University of Nice. "Everybody finds something. But trying to find an explanation for that is much more difficult."

Bias Abounds

The notion of implicit bias — that, thanks to the environment in which we’re raised, people can hold unconscious, unintentional biases against others — is an idea that’s at the center of much recent research about women in science.

It’s a concept easily grasped, and it jibes with the experience of many women in science, half of whom have reported experiencing gender bias, according to a 2010 survey. The idea of implicit bias also allows scientists to acknowledge that, even if they’re not overtly sexist — telling dirty jokes, or worse — there can still be a problem, said Ian M. Handley, an associate professor of psychology at Montana State University.

A number of individual studies of the role of gender in the sciences have had made headlines in recent years: In 2012, a study led by Corinne A. Moss-Racusin, now an assistant professor of psychology at Skidmore College, found that within a nationwide sample of 127 science professors, men and women alike pick a man over a women when asked to choose between two hypothetical, equally but ambiguously qualified undergraduate candidates to manage their lab.

Katherine L. Milkman, an associate professor of operations, information, and decisions at the University of Pennsylvania, led a study, published last year, in which more than 6,500 tenure-track professors at top research universities received emails. Posing as students, of varied sex and ethnicity, the researchers asked for mentorship. The professors, the study found, were most likely to say yes to white men.

And this year, in PLOS ONE, a study of a large introductory biology class at the University of Washington found that male students chronically underestimated the knowledge of their female peers in the class and overestimated the men’s.
"It kind of slapped me in the face," said Daniel Z. Grunspan, a graduate student in anthropology at Washington who, with Sarah L. Eddy, led the biology paper. This class is the students’ introduction to the field, he added. "It sets the tone for what they will or won’t do."

Experiments also show that the association of men with science appears to start from an early age, and lingers. Both male and female scientists hold it, but it’s especially potent in men, according to a recent review from a project that has tested more than 175,000 college-educated adults for bias. Appearance matters, too: A study published last month found that, presented with neutral pictures of real, working scientists, women judged to be more "feminine" were rated less likely to be scientists.

Bias is pernicious: It can even prevent its effects from being understood. Another experiment, published last year, discovered that male science faculty rated an abstract of an implicit-bias study less favorably than women did. And when the abstract was reversed, finding no bias, the female faculty rated it lower.

"We all do this," Mr. Handley said. "Let’s have a frank conversation about that."

Gains and Gaps

While these studies of bias have been fruitful, it remains unclear how they connect to the actual flow of women into science.

Women have made gains in the life sciences, where they receive some 70 percent of bachelor’s degrees, yet they hold only 38 percent of assistant professorships in the field, and only 24 percent of the full professorships. Underrepresentation is starker elsewhere: in physical science, technology, engineering, and math, it’s not uncommon for women to constitute a tenth of full professorships, and the ratio of American women earning doctorates in these fields peaked at 28 percent in 2009. It has been declining since.

While women do finish science doctorates as frequently as men do, there are problems at both ends of the pipeline: getting women into math and physical sciences as undergraduates and, at the conclusion, moving women with doctorates into tenure-track positions.

"Until young girls embrace advanced math in middle school and high school, they are less likely to major in math and math-intensive fields in college," said Wendy M. Williams, a professor of human development at Cornell University.

Women with doctorates, especially those in life science, psychology, and social science, are not making it to tenure-track positions at a proportionate rate, instead abandoning careers during the harsh, prolonged years of the postdoc, or finding part-time, lower-paid work. In these fields, from 2008 to 2010, women held 53 percent of the doctorates but made up only 32 percent of assistant professors.

Women who do persist through the postdoc may find academe a more welcoming home than these figures might suggest. One large recent survey of six scientific disciplines has found that, while women applied for tenure-track jobs in small numbers, those who did were more likely than men in the same applicant pool to get an interview and be offered a job. And once women got on the tenure track in science, a review by Ms. Williams and Stephen J. Ceci has shown that, as they put it, there was little difference over all between how much they and their male peers were paid and promoted, how long their careers lasted, or how satisfied they reported being in their jobs, though some persistent gaps remained, especially for pay in certain disciplines like economics and for hiring in the life sciences.
"Today," Ms. Williams said, "the playing field in academic science is mostly gender-fair."

How could this be true, given the high-profile bias studies?

When it comes to hiring and promotion, it’s possible both have part of the elephant. Some work has shown that excellent information about competence can mitigate bias, and faculty candidates have long records of objective information to assess, unlike potential lab managers. Or, if implicit biases remain influential, they could be countered by professors who have internalized the message that hiring women is vital to their discipline.

Disparities in Publishing

Women continue to face disadvantages in measures of publishing, which, in a time of metrics, can hold back their careers. They publish less than men, and a recent analysis by Ms. Sugimoto and others found that men receive roughly 10 percent more citations per paper than women.

One reason for this disparity is clear: Across academe, men are 50 percent more likely to cite their own work than women are. This dynamic, combined with an accounting for the inactive research years that may accompany family obligations, is enough to erase the productivity gap, a study in BioScience found earlier this year.

The lack of female mentors could be another cause of lower publishing rates, which is one reason Mr. Pezzoni looked recently at the relationships of Ph.D. advisers and graduate students at the California Institute of Technology. There was no difference in the publication rate of female students who had female mentors and the rate of their male counterparts who had male mentors. However, female students who had male mentors published 8.5 percent fewer papers than men advised by men.

Ms. Sugimoto’s new work on contributor roles could also help clear up differences in productivity, if they do exist, in biomedicine, said Molly M. King, a doctoral student in sociology at Stanford University. This division of labor could be driving the productivity gap, especially in fields where experiments are time intensive.

It would be wrong to immediately chalk up the finding to bias, Ms. Sugimoto said. It could be that women are more likely to micromanage a team. "And maybe that’s a better way to do science," she said, given that men seem more prone to committing fraud. But it’s also not clear how well these assigned credits actually correlate to the true breakdown of labor.

There’s much more work to do, of course. But in the meantime, Ms. Sugimoto said, scientists should put more thought into how they assign their work and how they train their teams to divide tasks.

Ms. Sugimoto’s study, and others like it, are the kind of subtle work that can greatly help the conversation about women in science, where discussions quickly get polarizing, said David I. Miller, a doctoral student in psychology at Northwestern University. "It’s been seen as either there’s pervasive bias against women that is everywhere, or that academic science isn’t sexist. But there’s a whole lot of interesting nuance in between."

Rather than asking whether gender bias exists, he continued, the more productive question is: "When does it exist?"

Paul Voosen is a senior reporter covering the sciences. Write him at paul.voosen@chronicle.com; follow him on Twitter @voooos; or see past work at voosen.me.