Hidden biases that drive women out of computer science

At the dawn of computing, women were early adopters of computational technology, using punch cards, which were considered the job of secretaries at the time. As computer science evolved into an honorary field focused on algorithms and theory, women remained dependent. Today, only 23% of Bachelor’s and PhD’s in Computer Science are awarded to women, and only 18% of Full Professors are women since the 1980s.

A new study by Dr. Samantha Kleinberg, Farber Chair of the Stevens Institute of Technology, reveals troubling patterns that help explain this persistent gap.

This work has been published in IEEE Access magazine.

Application vs. theoretical difference

Research in many fields generally falls into two categories. Applied Research aims to create new products, technologies, or solutions to specific real-world problems, such as developing algorithms to improve medical diagnosis and create systems that address social inequality.

In computing, theoretical research seeks to gain deeper insight into fundamental principles, such as proving the mathematical properties of algorithms and promoting understanding of computational complexity.

“When you enter the room at an applied computing meeting, you can see a balance between female and male attendees,” Kleinberg said. “In a conference focused on theory, the rooms look very different. There are significantly fewer women than men.”

While both types of research are essential to advance computer science, Kleinberg’s research reveals that they are not equally valued by the academic community. This may reflect a traditional academic preference for theoretical research that requires deep mathematical expertise, although many researchers have contributed to both fields throughout their careers.

This pattern reflects prior research showing that male-dominated subfields, such as computer science theory, tend to have higher institutional fame than domains represented by women, such as human computer interactions. Kleinberg’s work goes further by examining specific perceptions, fundraising decisions and citation patterns.

Revealing systematic bias

The disparity combined her personal experience with negative views of applied research prompted Kleinberg to conduct comprehensive research with Jesseca Marsh, a professor of psychology at Lehigh University.

They investigated tenure and tenure track faculty in the top 100 US computer science departments to understand the perceptions of researchers engaged in applied and theoretical research.

The findings reveal important biases against applied researchers and their research.

The faculty reviewed researchers who published their works in prestigious venues, who were rated as tenure or promotion, winning awards and less likely to earn funds. More concerning, faculty rated these researchers as more brilliant, creative and technically skilled than their theory-focused counterparts.

“I wanted to understand this dynamic that I was seeing,” explains Kleinberg. “So let’s find out what people actually think about this research and who do it.”

Data checks for bias

A comprehensive analysis confirmed the findings. Data from publications, employment, funding and awards show that applied research actually leads to worsening career outcomes.

Kleinberg subsequently used data from the authors of the publication to test the hypothesis that women are more representative in applied studies. Rather than using a tool to gender-match her first name to ensure the accuracy of her analysis, Kleinberg manually reviewed profiles of over 11,000 American scholars.

“I looked into all 11,524,” she shares. “There are tools that do it automatically based on names, but I had to do this manually because it’s not that accurate for Chinese names or other names with less gender.”

Kleinberg found that women are more highly representative in applied research fields than the theoretical ones. In other words, this bias disproportionately affects carrier outlook.

Recruitment Paradox

The irony is amazed. The university has successfully increased the participation of women in computer science by highlighting applications. When universities introduced interdisciplinary CS+X programs, the number of female students increased significantly, combining computing in areas such as anthropology, biology and music.

These programs appeal to students who want to apply their coding and algorithm construction skills, solving real-world problems rather than pursuing computing for itself.

“It’s not clear whether it’s actually their interest to make the theoretical work appealing, or whether it’s the culture of the field,” Kleinberg said. “Women may want to do theory, but may feel that they are not very welcome in those spaces.”

This study suggests that academia could drive women out of theoretical computing into the field of application through cultural barriers and fines for their work.

Why is this beyond academia?

Computer science benefits from a variety of perspectives and perspectives, and suffers when they are lacking. Just as early clinical trials that excluded women as subjects led to treatments that were less effective for women, computing research requires a diverse range of voices to create algorithms and tools that are useful to everyone.

“I’m studying health,” points out Kleinberg. “In the end, we hope that our algorithms and tools will be used for everyone and applied to everyone.

The implications of this study go beyond gender equity. As applied computing is already transforming healthcare, criminal justice and accessibility technologies, a systematic devaluation of this work could discourage critical research that addresses society’s most pressing challenges.

I’ll move forward

Kleinberg is similar to the way academic institutions underestimate their usual education and services compared to research. “It’s interesting to see the same disparity when it comes to theoretical and applied research. Scholars believe the work itself is worth doing, but it’s not that rewarded.”

Addressing this bias requires a systematic change in how universities assess their impact on research, how they train faculty in training to recognize unconscious bias and structure promotion and tenure decisions.

This study was conducted with approval from the Stevens Institute of Technology’s IRB. The study included a survey of 100 faculty members from the top-ranked computer science division and an analysis of publication, funding, and award data across multiple venues and programs.