The Female Factor: Is the gender gap in computer science carrying over to biomedical computing?
Early evidence suggests a mixed picture
Stanford University’s 2004-2005 computer science T-shirts exhibited symbols for six men and one woman -- an accurate portrayal of the ratio in the department and only slightly worse than the ratio nationwide. This country is facing a growing gender gap in computer science. While the percentage of women majoring in other science and engineering disciplines grew steadily in the past decade, computer science hovered at about 28 percent at four-year colleges (down from 37 percent in 1985) and 18 percent at PhD-granting universities (National Science Foundation; CRA Taulbee Survey).
In contrast, women’s representation in biology and medicine has soared, surpassing 50 percent in undergraduate biology in 1996 and in medical school in 2004—despite the strict male-domination of medicine just a generation ago (NSF; Association of American Medical Colleges). So what happens when biomedicine intersects with computer science? Many believe that biomedical computing has the potential to draw women into computing. It offers an antidote to computer science’s image problem: whereas pure computing is stereotyped as machine-oriented, solitary, and “geeky”, biomedical computing is seen as human-centered, team-oriented, and socially relevant.
But the early evidence shows a mixed picture. Biomedical Computation Review surveyed several universities with programs in biomedical informatics—Columbia, Harvard, the Massachusetts Institute of Technology, Stanford, the University of California, Los Angeles, and the University of Michigan—and found that, so far, just 24 percent of graduate students in these programs have been women. That’s closer to women’s representation in computer science than in biology.
Because biomedical computing has traditionally drawn from students with computational backgrounds, the deficit in females pursuing computer science as undergraduates may create a bottleneck. But a lack of women entering into the field is only half the problem. Across all scientific disciplines, including biology and medicine, the numbers of women shrink as you go up the academic ladder. This “leaky pipeline” phenomenon is worse in the more technical fields. Biomedical computing is no exception: Women make up just 13.8 percent of tenure-track faculty in biomedical informatics and computational biology at the surveyed institutions.
Biomedical Computation Review spoke to women in computing, biomedical computing, sociology, and education about why so few women go into computing; why they leave; and what biomedical computing can do to avoid the gender gap of computer science.
Their message: biomedical computing holds the promise of drawing more women into computing, but this influx may not happen spontaneously. Moreover, biology and medicine are becoming ever more computational. If women don’t have the computer savvy or the influence, they may be left out. More than being an injustice, they say, this would be a huge loss of talent and perspective for both biomedical computing and biomedicine.
PRECOLLEGE: NATURE OR NURTURE?
The gender gap in computer science starts early. Though equal numbers of men and women enter college with prior computer experience (UCLA Higher Education Research Institute) and young women are using computers at an incredible pace, only 14.8 percent of those taking the computer science AP exam in 2005 were women. Compare that to 46.3 percent for calculus (College Board AP Program).
Boys develop a “magnetic attraction” to computers at an early age, observe Jane Margolis, EdD, and Allan Fisher, PhD, in Unlocking the Clubhouse: Women in Computer Science (MIT Press, 2002). In their 1994-1999 interviews of 46 male and 51 female computer science majors at Carnegie Mellon University, three-quarters of the men—but only one-quarter of the women—fit the profile of someone who spent much of their youth consumed with computers.
“We found that while girls were not frightened of the computer or disinterested, there was a difference in how involved they got with computing,” says Margolis, a researcher at the UCLA Graduate School of Education.
Some attribute this disparity to innate gender differences: Women are nurturing and human-focused; men are analytical and object-focused; so, computers appeal more to men.
But socialization is working to a greater degree than we realize, argues Maria Charles, PhD, professor of sociology at the University of California, San Diego. “Both men and women believe in these fundamental gender differences. Independent of their truth, these beliefs can be very powerful in affecting people’s choices.”
In her study of 21 countries, she found that the male-to-female ratio in computer science (in college) varies widely: from 1.79 (in Turkey) to 6.42 (in the Czech Republic). This variation is better explained by societal and cultural norms than genetics. Similarly, the under-representation of minority men in computer science cannot be explained by innate gender differences.
Margolis also found that socialization plays a major role. In her interviews, 40 percent of the male students reported being given a computer early in life, compared with only 17 percent of the female students. “When parents place computers in boys’ bedrooms and spend more time nurturing their sons’ computing interests than their daughters’, are they responding to innate difference in the children’s level of interest? Or are their assumptions about the children’s interests playing out as self-fulfilling prophecies?” Margolis asks in Unlocking the Clubhouse.
Suzanna Lewis, MS, of the University of California at Berkeley, recalls the early influence of her four older brothers: “I think by the time I came along, I didn’t get treated any differently. Boys were who I played with. So it never occurred to me that being female meant that much.” She later designed control systems for steel mills and shipyards for a decade (which she likens to working on “a big tinker toy set”), before moving to bioinformatics. She is currently the only female out of 13 principal investigators involved in the NIH-funded National Centers for Biomedical Computing.
Grace Peng, PhD, grew up with math exercises over dinner. “I have a Chinese background. And so engineering was not thought of as something very strange for a woman,” she says. When she was 10 years old, she attended a program at the University of Illinois, Urbana-Champagne aimed at encouraging girls in computing—which she says made her comfortable with computers from an early age. She is now Program Director in the Division of Discovery Science and Technology at the National Institute of Biomedical Imaging and Bioengineering (NIBIB) managing programs related to computation and engineering systems.
If socialization is playing a major role, then early access and encouragement may be keys to effecting change. For example, foundations that donate computers to schools should require schools to demonstrate that girls are using the computers as much as boys—who tend to be more aggressive about grabbing them, says Cherri Pancake, PhD, Professor of Electrical Engineering and Computer Science at Oregon State University. Charles also urges parents and educators to downplay gender stereotypes and to require more math and computer science classes for everyone.
Finally, whether nurture-or nature-driven, girls tend to be less interested in the computer games that appeal to boys and more interested in how computing can solve real problems. Thus, integrating computing into subjects outside of the computer lab, such as biology, may help draw more women into technology before college.
COLLEGE: “GEEK CULTURE”
Tying computing into compelling problems—such as those in biology and health—may also help discredit the computer science stereotype of “this geeky white guy sitting behind a terminal, getting his suntan from the terminal rays,” says Lucy Sanders, MS, CEO of the National Center for Women & Information Technology (NCWIT) at the University of Colorado, Boulder. She formerly worked at Bell Labs for 24 years, including as a Chief Technology Officer.
“I think that it is an inaccurate and a repulsive kind of an image that keeps terrific men and women away. Not just women. Both,” she says.
“I don’t find the computer science world to be very socializing,” agrees Mia Levy, MD, a first-year student in the bioinformatics PhD program at Stanford. She says this stereotype was reinforced by her computer science classes, where the assignments were always to “create a computer game.”
“Having the month before gone from the critical care unit to then suddenly programming Boggle, I was like, ‘My life seems very insignificant right now in comparison to how important it all seemed before,’” she says.
Besides curriculum reform, broadening admissions criteria for computer science can also attract more diverse candidates, Sanders says. Requiring extensive computing know-how for freshmen is crazy, she says, and also irrelevant—since the technology world changes so quickly.
In response to Margolis’ study, the computer science department at Carnegie Mellon University changed their admissions imperative to finding thinkers and leaders—“and they got fabulous thinkers and leaders,” Sanders says. This admissions change, coupled with curriculum change, increased female representation from 7 to 42 percent in just six years. It also changed the computer science culture at Carnegie Mellon for the better, Margolis says.
Biomedical computing could draw from a large pool of thinkers and leaders in biology and medicine if they keep the technical barriers to entry low, Sanders says.
COLLEGE: THE CONFIDENCE GAP
Women often begin college with less computing experience than their male counterparts. “Boys have been tinkering and experimenting and working at the computer since they were very, very young; it becomes almost a physical intelligence,” Margolis says. This may intimidate women, and erode their confidence at a time when—studies show—women generally experience a dip in confidence.
Indeed, Pancake found that women were leaving the computer science major at Oregon State University because they believed they couldn’t keep up—but, in fact, they were doing just as well as the men.
“What we believe from observing in the classroom is just that men tend to act more self-assured, not that they’re any better prepared,” Pancake says.
“It’s the imposter syndrome,” echoes Francine Berman, PhD, Director of the San Diego Supercomputer Center. “People of both genders feel like secretly they don’t have what it takes, and they’re just trying to make sure that no one knows that. Women tend to take it more seriously and personally. They are encouraged not to stick with it, because they’re afraid it’s true. But it’s not true.”
Interestingly, Margolis found that international women seem to better resist this confidence trap; though they often had no computing experience to start, they stuck out the rigorous program at Carnegie Mellon University. International students expressed the belief that if you work hard, you will make it, but American women too often succumbed to the belief that intelligence is fixed--not malleable--and either you have that certain intelligence or you don't.
“There have been studies showing that if you believe that intelligence is fixed and innate, your confidence is much lower than if you believe that, with work, it's malleable,” she says.
Confidence and competence are not always related, Sanders advises. Rather than accepting confidence at face value, she developed a trick: “Somebody would say something like they just knew it was true—which is common in my discipline—and I would say: ‘oh, why?’ I’d go through two or three rounds of questions before I would decide whether they knew what they were talking about or they didn’t.”
Tinkering with computers doesn’t make you a good computer scientist, adds Samantha Chui, a Stanford University computer science graduate who is now a first year master’s student in bioinformatics: “I feel like there are two kinds of people: the people who know everything about the hardware and can’t program, and the people who know how to program but don’t know how to fix the computer. So I just categorize myself as the latter, and move on.”
GRADUATE SCHOOL: LEAKY PIPELINE
Women who enter graduate programs in science are less likely than their male counterparts to complete a PhD and to move on to tenure-track faculty positions, particularly at top research institutions. A year ago, Harvard University President Larry Summers, PhD, ignited a firestorm with his provocative suggestion that innate differences in ability play a role. It's an idea uniformly rejected by our panelists. They say women are opting out because they perceive the academic environment to be unappealing (more specifically: unwelcoming, harsh, individualistic, combative, competitive, political, secretive, and focused on empire building) and incompatible with having a life or raising a family.
These impressions are formed early in graduate school and they’re largely untrue, says Semahat Demir, PhD, associate professor of biomedical engineering at the University of Memphis and the University of Tennessee. “When graduate students see their faculty as role models, they don’t get a good feeling for how much the faculty enjoy their jobs,” she says. “They see the faculty working so many plus hours. They see deadlines and frustrations.” But, in fact, she says, academia is very fulfilling and offers a lot of flexibility for balancing life and work.
“Women faculty have to talk to these women early and show that you can get a balance,” she says. She advocates mentoring to facilitate communication between faculty and students, and to help women students develop soft skills that are otherwise inadequately addressed in graduate school, such as negotiation.
“I know for myself, if I’m in a kill ‘em and eat ‘em environment, that doesn’t encourage me to do my best work,” says Berman, who has promoted a more collaborative environment at the San Diego Supercomputer Center. Though she doesn’t make a conscious effort to recruit women, they make up half of her senior management team.
Biomedical computing is team-based and collaborative by nature, which may help to attract and retain women, our panelists say.
FACULTY: SECRET LIVES
After earning a science doctorate, women are more likely to leave academia or to take a non-tenure track position—choices often attributed to a conflict between the biological clock and the tenure clock. Among science and engineering doctorates who are not working, 35 percent of women cite family responsibilities as the reason, compared with only 2 percent of men. Among employed doctorate holders, 33 percent of women versus just 17 percent of men have never been married (NSF).
“Professional life in general and high stakes professional life in particular is not family friendly, for either women or men. You’re incentivized for working all the time and for choosing your career over your family,” Berman says. “Professional success encourages you not to have a life. But everyone does have a life, so they have secret lives,” she says.
Berman now makes a point of talking about her family, to show young women that it is possible to have both a family and a successful career in computer science. Academia actually has a lot of flexibility for parents, she says. Earlier in her career, she says, “If I had to go home to watch my daughter play a fork in the school play, I’d just say I had an off campus commitment. My husband did the same.”
FACULTY: SUBTLE EXCLUSION
Even when women decide to pursue a tenure-track academic position, they may face unconscious biases in hiring, compensation, and evaluation. For example, studies show that both men and women will rate an identical resume or journal article lower (on average) if they are told that the subject or author is a woman. A 1999 report on the status of women in science at MIT revealed that, even when controlling for other factors, women faculty members received less pay and less lab space than men—findings that were replicated at other elite institutions.
Jeanette Schmidt, PhD, executive director of Simbios at Stanford and former Vice President of Research and Development InCyte Corporation, says she never felt her capabilities doubted as a graduate student in Israel. But as an assistant professor at Polytechnic University in New York, she initially felt a subtle exclusion and lack of respect that she attributed to being a woman.
The manifestations were often well meaning. For example, she was assigned to teach a morning undergraduate course rather than a more desirable evening graduate course, because—she was told—she should “be home with her kids at night.”
Biases creep into academia more than industry, Lewis speculates. In industry, you are judged by how you affect the bottom line, she says. In academia, you are judged on something less tangible—your ability to debate, persuade, and make yourself known. These kind of subjective criteria introduce a soft bias that favors the dominant group, a “deafness” as she describes it: “How many times have you [as a woman] been standing in a group and you say something and nobody recognizes it? And then someone else who happens to be male says the same thing, and everybody says ‘ah, what a great idea!’”
Counteracting unconscious biases requires a pro-active approach. By actively recruiting women, MIT increased from 22 to 34 women faculty in the School of Science from 1994 to 1999 (previously, the faculty had stuck fast at 8 percent women for more than a decade). At Oregon State University, Pancake says that sending targeted recruitment letters to eligible women and minorities has increased their proportion of women engineering faculty to the fourth highest in the country.
Biomedical computing needs to be similarly pro-active in recruiting and promoting women, our panelists stress. And this impetus can’t just come from women in the field.
"Typically you see a lot of the programs developed and executed by women for women. But it's not a women's problem, it's everybody's problem," Berman says. "If it was important enough for everyone to address this problem, it would cease to be a problem."
TWO RIVERS MERGING
Molecular biology and computing were born in the same era and have grown up side by side, like two rivers winding down a mountain in parallel, Lewis reflects. The two rivers are now merging, as computing becomes more and more integral to biology, she says. And before too long, it’s not going to be easy to distinguish between the two fields.
Computing is becoming integral to many other disciplines as well, which is precisely what makes the gender gap in computer science so worrisome. At a time when too few U.S. students are going into technology to keep up with the demand, we can’t afford to lose half our talent pool.
We also can’t afford to lose women’s perspectives, Sanders says. “Technology can’t be designed and invented by a homogenous group of people because it will be less than it should be.”
Biomedical computing is poised in a unique position to open up computing and technology to a larger audience, in particular women. But it’s still unclear if this promise will be realized.
In 2004-2005, only 1 in 7 computer science majors at Stanford was a woman, but 2 in 7 biomedical computation majors were women. Whether or not this percentage will rise or fall is uncertain, but which way it goes will impact the future of several disciplines: not only biomedical computing, but also biology, medicine, and computer science.