Over the last three years, Sarah Eddy and Daniel Grunspan have asked over 1,700 biology undergraduates at the University of Washington to name classmates whom they thought were “strong in their understanding of classroom material.” The results were worrying but predictable. The male students underestimated their female peers, over-nominating other men over better-performing women.

Put it this way: To the men in these classes, a woman would need to get an A to get the same prestige as a man getting a B.

“A lot of people make the assumption that issues of gender in biology are gone because so many women enroll,” says Eddy. “But we know there are strong unconscious biases equating science to males. They’re just there in the air.”

Her study is the latest to show the challenges faced by women in science, technology, engineering, and math (STEM). In the U.S., women earn around half the doctorates in these fields, but so many drop out at every step of the career ladder, that men always dominate the top echelons. As Helen Shen writes in Nature, women comprise “only 21 percent of full science professors and 5 percent of full engineering professors” and “on average, they earn just 82 percent of what male scientists make in the United States—even less in Europe.”

The causes of this attrition are manifold, but sexual discrimination is an indisputable part of it. Women in STEM repeatedly report experiencing sexual harassment, being mistaken for administrative staff, being forced to prove themselves to a degree that their male colleagues are not, being told to behave in more aggressive, outspoken masculine ways while simultaneously facing backlash for doing so.

And several careful experiments have shown that faculty members—both men and women—are more likely to spend their time mentoring men, to respond to emails from men, to call on men in classes, to rate (fictional) male applicants as more competent and hirable than identical female ones, and to hire a man for a job that requires math.

These biases, sometimes manifesting outrightly and sometimes insidiously, collectively create an environment where women feel like they don’t belong, like they aren’t valued, like the odds are set against them. Confidence falls, perseverance wanes, and careers die by a thousand cuts.

It begins early. Eddy has been studying the University of Washington’s undergraduate biology course for a few years to try and understand how biases play out among the students themselves. She teamed up with Daniel Grunspan, an anthropologist who’s interested in how information travels within groups. They surveyed three large classes of 196, 759, and 760 students respectively, asking them to nominate particularly strong peers at various points through the academic year. They found that men consistently received more nominations than women, and this bias only got worse as the year went on. The question is: Why?

Performance? Men got better grades than women in the three classes but the difference was only statistically significant in one; even then, the scores differed by no more than 0.2 of a grade-point average. Participation? The class instructors deemed more men than women to be “outspoken,” and Eddy’s previous work certainly showed that women comprise 60 percent of the students, but just 40 percent of the voices heard in class.

But even after adjusting for both these factors, the team found that male students still disproportionately nominated other men, giving them a boost equivalent to a GPA increase of 0.77. By contrast, the female students showed no such biases, giving other women a paltry boost of just 0.04 GPA points. As the team wrote, “On this scale, the male nominators’ gender bias is 19 times the size of the female nominators’.”

The team also found that the ‘celebrities’—the three students in each class with the most nominations—were all men. Sure, they had good grades and spoke up frequently, but they all had female peers who were equally outspoken, with grades just as high. As Grunspan and Eddy wrote, “It appears that being male is a prerequisite for students to achieve celebrity status within these classrooms.”

“They have the right of it,” says Kate Clancy from the University of Illinois at Urbana-Champaign. “Their paper is consistent with the ways in which implicit bias influences who we tend to see as a scientist—if we culturally associate maleness with scientific abilities, it makes sense that we'd overvalue men's contributions in the science classroom.”

“It's also pretty consistent with the natural experiment I've been in for the past 10 years as a female scientist married to a male scientist,” she adds. “The junior female faculty that I've started mentoring in recent years report the same thing: They have to beg and plead and buy coffee for colleagues a million times before anyone associates their expertise with their name.”

Eddy expects that even stronger biases lurk in other STEM fields. After all, there are even stronger negative stereotypes about female ability in physics, maths, and engineering. And in these subjects, women are typically outnumbered in classes. They must contend not only with the same biases that biology student face, but also with stereotype threat—a well-documented phenomenon where the anxiety of fulfilling a negative stereotype hampers the performances of people from minority groups.

But Eddy takes it as a hopeful sign that the women in the study didn’t show biases towards their female peers, especially since other researchers have found that gender biases exist among female faculty members. “It’s hopeful,” she says. “Maybe things are changing culturally, helping women to overcome those historical biases.”

She has also tested some psychological tricks that have helped students to cope with stereotype threat in past trials, including simple writing exercises designed to combat stereotype threat by affirming a student’s values. Other “band-aid solutions” might help too, including doing more work in small groups where women feel more comfortable participating, or having more female role models up front. The instructors in the classes that Eddy studied were almost all men, and it’s perhaps no coincidence that the one with the lone female instructor also had the smallest gender biases.