Maria Klawe has succeeded where many other leaders have failed.

At Harvey Mudd College, the Claremont, California-based institution where Klawe became president in 2006, a full half of students who graduate with degrees in computer science, engineering, and physics are women. In comparison, just 18 percent of computer science graduates nationwide are women--a percentage that's been shrinking since 1985, when it stood at 37 percent.

In a recent speech at IEEE's Women in Engineering International Leadership Conference, Klawe detailed just how Harvey Mudd reached this milestone, how the campus has changed in response, and, perhaps most soberingly, why so few institutions have followed in her footsteps.

Klawe says that Harvey Mudd has been mostly male and white since its founding in 1955. Until 1971, the share of students who could be female was capped at 11 percent. It's long been an elite school, competing for students with bigger sciences powerhouses such as Cal-Tech and MIT. In 1996, says Klawe, female students made up 20 percent of the student body. Ten years later, when Klawe became president, that number stood at 30 percent. Now, half the student body is female.

In 2012, about two percent of Harvey Mudd students were African-American. That number is now closer to 10 percent. Hispanic students make up about 20 percent of the student body, up from five percent. Native American and Pacific Islanders are about three percent, up from one percent. The Asian American population at the school has remained stable at about 20 to 25 percent.

Faculty, Klawe noted, tend to stick around a lot longer than students, so faculty demographics have been slower to change. The faculty is currently 40 percent women--whereas "Cal-Tech recently made it into double digits," said Klawe--and six out of seven department chairs are female. Klawe is the college's first female president.

Three Steps to Diversity

"I don't think when Harvey Mudd hired me they really understood how committed I was to diversity," said Klawe, who had previously been the dean of engineering at Princeton University. When she first started talking about how the school needed to become more diverse, she says, "I was surprised to find out the meme around Harvey Mudd was that we are a merit-based institution and bringing in more women or people of color would mean lowering our standards." The first conversations about diversity, she said, "were not very encouraging."

But then, at her third strategic planning workshop, Mudd brought in Freeman Hrabowski III, the president of the University of Maryland , Baltimore County. "Freeman is extraordinary," said Klawe. "After a 20-minute talk, the language just changed. Unsurpassed excellence and diversity at all levels became part of the strategic plan." Importantly, that priority was embraced across the Harvey Mudd community, including faculty, staff, students, alumni and trustees, paving the way for its success.

Klawe started with the hypothesis that if Harvey Mudd created an environment that was supportive and engaging for everyone; if the school built confidence and community among underrepresented groups; and if it demystified the path to success, a diverse group of students would be attracted to the college and succeed there. In computer science, engineering, and physics, said Klawe, "underrepresented" means women and people of color. Klawe noted that she thinks this three-part hypothesis is true of workplaces, as well.

The computer science department had been well aware of the diversity problem before Klawe's arrival, and had redesigned its introductory class. One of the keys was to separate sections in that class by experience. "If you want to encourage someone in the introductory course, you don't put them in the course with people who have a lot more experience and don't look like them," she says.

One section of the class was for people who really were beginners and hadn't programmed before. Another section was for people who had some experience. A third section, called "green," was for beginners, but all the applications in that section have to do with biology. The last section was for people who really shouldn't have been in an introductory class at all, so they took the first two computer science classes together.

One of the keys to making this work, said Klawe, was acknowledging that three of these sections would all need to be on the same footing when they started the second computer science class. So she made sure that none of the advanced material included in the sections for more-experienced students would be at all helpful in the second class. "It's interesting and useful, but not until you're a junior," she says.

Next, professors were to set the expectation that success in the class was dependent on hard work and asking for help. "In disciplines that say it's about innate ability, there are fewer women," said Klawe. Problems were to be framed as creative problem solving, using real-life examples. Professors encouraged collaboration and paired students on homework assignments. "The worst thing is to think you're the only one, there's no one to ask, and you can't do it," said Klawe.

The professors also endeavored to eliminate what Klawe referred to as macho behavior. There are students, she said, who love the course but dominate discussion with their constant comments and questions. Professors should tell these students, in private, something like, "I love having you in my class, and I love your enthusiasm for the material. Would you mind if we had these conversations during office hours? A lot of the other students are somewhat intimidated."

Last, Klawe made sure all the students were encouraged to take the second computer science course. "The vast majority of women who show up at Harvey Mudd didn't expect to want to major in computer science," said Klawe. "By the third course, they're like, 'I could do this.' Because the classes are 50/50 [male/female] it doesn't seem weird to be a female in computer science, engineering, or physics."

As for why more schools haven't followed her lead, Klawe said she didn't exactly know the answer. But she hinted that  arrogance may be part of it. "In some disciplines, such as physics, there's still an enormous attachment to the idea that talent in physics is something you're born with. That intellectual firepower and the ability to reason and argue on your feet is part of being a great physicist," said Klawe. "Well it's nonsense. Everything that you learn in your life can be improved with practice."