The STEM Gender Gap Isn’t Where You Think It Is

Imagine two university students. Both are women. Both are studying subjects commonly associated with STEM. One studies Engineering and technology. The other studies Veterinary sciences.

In 2023–24, women represented 19.1% of all undergraduate students in Engineering and technology at English higher-education providers. In Veterinary sciences, they represented 84.8%.

That is almost a mirror image — and it exposes a problem with one of the most familiar phrases in discussions about education and equality: “the STEM gender gap.”

There is no single STEM gender gap. There are several.

THE AVERAGE THAT HIDES THE STORY

STEM is a useful acronym. It allows governments, universities, employers and educators to talk collectively about science, technology, engineering and mathematics. But useful categories can become misleading when they make very different subjects look alike.

Computing and Psychology do not have the same gender balance. Neither do Engineering and Medicine, nor Mathematics and Veterinary sciences.

She Invented That analysed Office for Students population data on female representation among all undergraduate students across ten selected subject groups at English higher-education providers between 2010–11 and 2023–24.

The result is not one trajectory. It is a collection of very different ones.

TWO WORLDS UNDER THE SAME UMBRELLA

In 2023–24, the female share was 84.8% in Veterinary sciences, 82.2% in Psychology, 79.6% in Subjects allied to medicine, 62.2% in Medicine and dentistry, 52.6% in Geography, earth and environmental studies, 49.3% in Biological and sport sciences, 45.8% in Physical sciences, 34.5% in Mathematical sciences, 20.1% in Computing and 19.1% in Engineering and technology.

At one end are Computing and Engineering, where approximately four out of five undergraduate students were not female. At the other are Psychology and Veterinary sciences, where more than four out of five were female.

The question, then, is not simply why women are underrepresented in STEM. It is why disciplines grouped beneath the same broad label can develop such radically different gender compositions.

PERCENTAGES ARE ONLY HALF THE STORY

A percentage tells us about the composition of a student population. It does not, by itself, tell us whether the number of women studying the subject is rising or falling.

In 2023–24, approximately 21,740 female undergraduates were studying Computing. They nevertheless represented only 20.1% of Computing students.

Both statements can be true because the denominator matters. If the number of women rises while the total student population rises even faster, female participation can grow substantially without producing an equally dramatic increase in female share.

For that reason, the She Invented That Statistics Observatory presents both female student numbers and female proportions rather than treating either measure as sufficient on its own.

COMPUTING IS CHANGING — SLOWLY

In 2010–11, women represented 16.5% of all undergraduate Computing students in the data. By 2023–24, that figure had risen to 20.1%.

Over thirteen academic-year intervals, female representation increased by 3.6 percentage points. The direction is positive; the pace is less impressive.

Computing therefore illustrates both progress and persistence. Women are participating in considerable numbers, but the overall population remains heavily male.

ENGINEERING MOVED FURTHER — AND STILL ENDED LOWER

Engineering and technology increased from 14.4% female representation in 2010–11 to 19.1% in 2023–24 — a rise of 4.7 percentage points.

In 2023–24, approximately 21,640 female undergraduate students were studying Engineering and technology. Yet Engineering still finished the period with the lowest female representation among the ten subject groups examined here.

A field can recruit a substantial number of women, improve its female share and remain deeply imbalanced at the same time. Engineering has changed. It has not transformed.

MATHEMATICS BREAKS THE SIMPLE PROGRESS STORY

Mathematical sciences complicate the picture further. Female representation was 38.7% in 2010–11 and 34.5% in 2023–24. Approximately 9,870 female undergraduates were studying Mathematical sciences in the latest year.

That does not by itself tell us why. A new subject coding system — the Higher Education Classification of Subjects, or HECoS — was introduced in 2019–20, replacing the Joint Academic Coding System, or JACS.

The Office for Students maps subjects from both systems into a Common Aggregation Hierarchy to support long-term comparison, but warns that the transition can create additional variation in subject distributions between 2018–19 and 2019–20.

It would therefore be unsafe to attribute Mathematics’ overall fall of 4.2 percentage points to reclassification. The coding transition may affect part of the year-to-year movement, but it does not by itself explain the thirteen-year trajectory.

Good data analysis does not make uncertainty disappear. It identifies where it matters.

PHYSICAL SCIENCES TELL ANOTHER STORY

Physical sciences moved from 38.6% female representation in 2010–11 to 45.8% in 2023–24, an increase of 7.2 percentage points. Approximately 19,070 female undergraduates were studying Physical sciences in the latest year.

Geography, earth and environmental studies moved from 46.5% to 52.6%, while Biological and sport sciences reached 49.3%.

If women were broadly deterred simply by scientific content, quantitative work or technical study, we might expect similar patterns across scientific disciplines. We do not see them.

THEN THE GAP REVERSES

Move into Medicine, health-related subjects, Psychology and Veterinary sciences and the familiar gender-gap narrative reverses.

Women represented 62.2% of undergraduate students in Medicine and dentistry in 2023–24, 79.6% in Subjects allied to medicine, 82.2% in Psychology and 84.8% in Veterinary sciences.

Nor is this entirely new. In 2010–11, female representation was already 79.7% in Psychology, 81.0% in Subjects allied to medicine and 81.2% in Veterinary sciences.

This does not make concerns about women’s participation in Computing or Engineering less important. It makes the underlying question more precise.

TEN SUBJECTS, TEN TRAJECTORIES

Between 2010–11 and 2023–24, female representation changed by +7.2 percentage points in Physical sciences, +6.1 in Geography, earth and environmental studies, +5.6 in Medicine and dentistry, +4.7 in Engineering and technology, +3.6 in Computing, +3.6 in Biological and sport sciences, +3.6 in Veterinary sciences, +2.5 in Psychology, -1.4 in Subjects allied to medicine and -4.2 in Mathematical sciences.

These figures measure changes in composition, not necessarily changes in absolute female student numbers. Some disciplines became considerably more balanced. Some moved slowly. Some were already strongly female. Some moved away from an even split.

There is no common STEM trajectory.

WHY DOES ENGINEERING LOOK SO DIFFERENT?

Population statistics can identify these differences. They cannot establish all of their causes. Other research, however, shows why discipline-specific pathways matter.

EngineeringUK research on progression from A levels into Engineering and technology found a substantial gender difference even among students who had studied Maths and/or Physics: 23% of men progressed to Engineering and technology degrees compared with 8% of women.

The Engineering imbalance therefore does not suddenly appear when university applications are submitted. Two questions matter: who reaches the relevant school subjects, and among those who do, who sees Engineering as the next step?

COMPUTING HAS ITS OWN ECOSYSTEM

Computing presents another mechanism. Its history shows that programming was not always culturally understood as male work. Women were prominent in many early forms of programming before computing became increasingly professionalised and culturally masculine during the second half of the twentieth century.

Research from the Gender Balance in Computing programme has also examined how attitudes towards Computing diverge during school and whether classroom context, pedagogy and perceptions of the subject can influence girls’ engagement.

The important point is not that one intervention explains the university statistics. It is that Computing’s gender imbalance has its own educational and cultural pathway.

STEM IS NOT ONE LEAKING PIPELINE

Discussions about women in STEM often use the image of a pipeline: students enter education at one end and scientists, engineers and technologists emerge at the other. If women are missing from the workforce, researchers search for the point at which they “leak” from the pipeline.

The metaphor has been useful because it encourages us to look beyond final workforce numbers. But the subject data reveal its limitation. There is no single pipe.

STEM is better understood as a collection of field-specific ecosystems, each with its own school pathways, admissions patterns, professional identities, labour markets, institutional histories and cultural expectations.

Instead of asking where women leak out of “the STEM pipeline”, we should ask what happens inside each disciplinary ecosystem: where an imbalance first appears, whether it grows or shrinks during education, who enters, who stays and who progresses.

BALANCE IS NOT THE SAME AS EQUALITY

A roughly equal number of male and female students does not prove that a discipline has achieved gender equality. Student representation tells us who is studying a subject; it does not tell us whether those students have identical experiences, completion rates, postgraduate opportunities, salaries, promotion prospects or access to leadership.

Nor does high female representation automatically mean that women dominate the most senior or prestigious parts of a profession. Education is one stage of a much longer system.

WHAT SHOULD CHANGE?

If there is no single STEM gender gap, there cannot be a single intervention capable of solving it. That does not mean broad initiatives supporting women in science and technology have no value. It means broad targets are not enough.

For policymakers, universities, employers and professional organisations, the first implication is simple: measure disciplines separately. An overall improvement in women’s STEM participation can coexist with very little movement in Computing or Engineering.

Second, interventions should be designed around the stage at which the imbalance develops. If a subject loses potential students before university, university recruitment campaigns alone arrive too late. If women enter in substantial numbers but disappear later from senior employment, the problem lies somewhere else again.

Third, success should be measured by outcomes rather than participation campaigns themselves: subject choices, applications, entry, retention and progression.

Finally, the most severely imbalanced disciplines deserve specific attention. In the 2023–24 data examined here, Engineering and technology and Computing stand apart, with female representation around one-fifth in both.

WHAT THE DATA DO NOT TELL US

These figures cover students at English higher-education providers, not every student in the United Kingdom. The analysis uses all undergraduate students and all domiciles in the selected subject categories.

The ten disciplines were selected by She Invented That for exploration; they should not be interpreted as an official Office for Students definition of STEM.

The transition from JACS to HECoS in 2019–20 also requires caution when comparing subject distributions immediately before and after that year. Population data reveal patterns rather than automatically explaining their causes.

The Observatory tells us what happened. Understanding why requires another layer of evidence.

THE QUESTION WE SHOULD BE ASKING

For years, a familiar question has shaped discussions about gender and science: Why are there so few women in STEM? The data suggest that this question is too broad.

There are not “so few women” everywhere. In some disciplines, women are a small minority. In others, representation is close to even. In several, women form a substantial majority.

More useful questions are discipline-specific: Why has Physical sciences moved substantially closer to parity while Computing remains around one-fifth female? Why has Engineering improved but remained so heavily male? What lies behind the long-term decline in the female share of Mathematical sciences? Why are Psychology and Veterinary sciences so strongly female?

Those questions are harder. They are also much more likely to produce useful answers.

THERE IS NO SINGLE STEM GENDER GAP

In the same higher-education system, at the same point in time, women can represent around one-fifth of students in one scientific or technical discipline and more than four-fifths in another.

Over fourteen years, one field can move steadily towards parity while another moves in the opposite direction.

Gender representation in science is not determined simply by whether a subject is “STEM”. It is shaped discipline by discipline — by education, history, culture, institutions, professional expectations and the choices people encounter long before they arrive at university.

The most revealing number may not be the percentage of women in STEM. It may be the difference between the percentages.

Because hidden inside a single acronym are several very different worlds.

Explore the data

Use the She Invented That Statistics Observatory to compare female student numbers and representation across ten selected undergraduate subject groups at English higher-education providers from 2010–11 to 2023–24.