“Engineering requires culturally masculine traits that both men and women have. The women in our study were as likely as the men to be interested in working with things, on average. But engineering also requires working with people to be effective — communication with the team, clients, regulatory bodies, etc. If women happen to be culturally socialized into enjoying that or having skills there, that’s a reason to steer them into engineering, not out of it.”
— Mary Blair-Loy, Ph.D.
Inherent data biases in career assessment tools, or CATs, used for advising students every year on college majors and careers can systematically exclude girls from certain career pathways, according to a November 2024 report published in the Sociological Inquiry. The research, published in the paper “Steering Women out of Engineering: Career Assessment Tools as a Technology of Self-Expressive Segregation,” highlights the importance of addressing data and algorithmic biases in such tests to promote equitable career guidance toward engineering.
The report examines how CATs, which are trusted by millions and considered an objective mechanism, are less likely to recommend engineering occupations to women, even after controlling for GPA, satisfaction with the major, and planned persistence. The CATs use small differences in students’ preferences when responding to test questions to drive them toward different occupation recommendations, exacerbating gender segregation in certain occupations and reinforcing men’s dominance in engineering, the report finds.
“Engineering and other STEM fields are interesting because people are committed to objective, evidence-based evaluation and a meritocracy that is fair and rewards the best and brightest. Researchers have shown in a number of specific ways within engineering that those expectations of objectivity and meritocracy are violated,” says report author Mary Blair-Loy, Ph.D., professor at University of California San Diego’s department of sociology.
The research considers two CATs that are widely used in educational institutions — O*NET Interest Profiler and Traitify Career Discovery. O*NET is freely available and based on the work of John Holland, Ph.D., who developed the original CAT in the 1950s. Dr. Holland’s RIASEC Interest Structure sorts people into six personality types: realistic, investigative, artistic, social, entrepreneurial, and conventional. Test takers rate how much they like or dislike various occupational tasks and these ratings are combined to assign the individual a primary and secondary personality type. Previous research has found that CATs based on Holland’s RIASEC system tend to disproportionately place women into working-with-people categories and men into working-with-things categories. And often, engineering is seen only as a working-with-things profession.
Traitify is a newer competitor in CATs, developed in 2011, which aims to identify seven personality types based on a broad range of psychological approaches. It then links those types to occupations. The test has 56 randomly appearing personality tasks or traits accompanied by corresponding photographic images. Test takers indicate whether they identify with the task or trait by choosing “me” or “not me.”
When using either option, test takers do not provide information about their gender or other demographics, or their current college or work experiences.
In Dr. Blair-Loy’s research, women participants overall had GPAs, scores on their satisfaction with their major, and plans to persist in their major that were comparable to men’s. But the women were far more likely than men to receive CAT recommendations that would steer them away from engineering careers (see the chart below). Using both CATs, women scored higher on the people-oriented personality type than men and therefore received significantly fewer engineering recommendations than men. Small differences in orientation between men and women are exaggerated by the CATs, resulting in women being recommended for engineering occupations at a rate of just one-third to two-thirds the rate for men, according to the study.
“It’s important to look at the broader picture of attrition among talented girls and young women who have expressed interest in engineering at younger ages,” says Dr. Blair-Loy. “Significant junctures at which women leave the field include when selecting a college major. And women who do decide to major in engineering are more likely than men to exit. So, college is an important juncture in the leaky pipeline. Engineering requires culturally masculine traits that both men and women have. The women in our study were as likely as the men to be interested in working with things, on average. But engineering also requires working with people to be effective — communication with the team, clients, regulatory bodies, etc. If women happen to be culturally socialized into enjoying that or having skills there, that’s a reason to steer them into engineering, not out of it. Unfortunately, CATs seem to use ‘working with people’ interests to steer many women out, even when they are succeeding in the major.”
CATs appear to be part of a broader process, also found in computer-assisted, semi-automated recruitment, in which algorithms are influenced by subtle gender cues and can reinforce biases present in the data on which they were trained, the study says.
Additionally, CAT instruments are embedded into facets of social life beyond advising students, the article states. O*NET assigns RIASEC codes to the more than 500 occupational titles delineated by the United States Department of Labor. And Traitify is used to provide input for dating apps, credit scoring, and other applications.
“One thing Traitify is used for is to determine [for] potential employers whether an applicant for a job seems to be a good fit. And there’s a lot of research on how, [in] a field numerically dominated by men and culturally masculine, like engineering, women are more likely than men to be not seen as a good fit,” Dr. Blair-Loy says. “CATs’ gender bias makes this nebulous ‘fit’ idea seem more scientific and more justifiable [as a reason] to exclude people.”
The report recommends using CATs as part of integrative career counseling combined with student-centered, subjective counseling. High school and college counselors should be made aware of how infrequently engineering is recommended to women so this bias can be accounted for, the report says.
“Students need to be informed about possible biases in CATs,” says Dr. Blair-Loy. “I’d like students to think more analytically and critically about the tests and their results and not take them at face value. That’s a real issue; how do we get this information out directly to students?”
