Women make up only 35 per cent of STEM graduates globally. Andrés Barrios Fernández and Aspasia Bizopoulou show that while there is huge variety between countries in the educational attainment of men and women, young women’s choices about studying STEM subjects look remarkably similar everywhere.
Around the world, women remain underrepresented in science, technology, engineering and mathematics (STEM). Globally, women make up just 35 per cent of STEM graduates – a figure that has barely changed in a decade. This headline number raises a fundamental question: why does the gap persist, even as girls’ overall educational achievement rises?
Understanding this requires going beyond enrolment statistics. Enrolment reflects both who applies to STEM programmes and who gets admitted, combining academic preparation, preferences and potential discrimination in the admissions process. To design effective policy, we must disentangle these components.
The pipeline and choice gaps
This is precisely what we do in new research. We bring together administrative data from ten centralised university admissions systems across five continents – Australia, Brazil, Chile, China, Finland, Greece, Spain, Sweden, Taiwan and Uganda.
These systems allocate places based only ontest scores and students’ stated choices, leaving no scope for discrimination in admissions. This design is powerful: it makes it possible to isolate two distinct components of the STEM gender gap.
First, there is a “pipeline gap”. This refers to whether women and men differ in reaching the academic threshold needed to qualify for competitive STEM programmes. Second, there is a “choice gap”, namely whether academically similar women and men – who in the contexts we study have the same admission probabilities – differ in ranking a STEM degree as their top choice.
Understanding these two components separately is essential. Policies that improve academic preparation target the pipeline. Policies that influence student choices target the choice gap. Without distinguishing between them, we risk aiming at the wrong target.
A unique global dataset on student choices
Centralised admissions systems give us something that most data sources cannot: students’ ranked lists of applications, not just their final enrolments. Because these systems use variants of the deferred-acceptance algorithm, students have strong incentives to report their true preferences. And because eligibility is determined mechanically by exam scores, students with similar scores face identical admission prospects.
We are essentially able to observe what students want to study, not just where they end up. Across the ten countries, we focus on the top 10 per cent of exam scorers – the students most likely to qualify for selective STEM programmes. For this group, we measure the pipeline gap (who reaches the top academically) and the choice gap (who selects STEM first).
This cross-country comparison has not been possible before. Prior work has examined academic performance or STEM enrolment within individual countries, but studying STEM choices directly requires application data, which are rarely available and comparable across such diverse settings.
Large global differences in the academic pipeline
The pipeline looks very different depending on where students grow up. In Uganda,only40 per cent of top-scoring students are women. In Chile, Brazil and Taiwan, young women remain underrepresented among high achievers. Yet In Finland, Spain, Australia, Greece and Sweden, the pipeline flips: women outnumber men among the highest performers. In Sweden, nearly65 per cent of top scorers are women.
These differences align strongly with broader measures of gender equality. Countries with higher gender parity tend to have more young women reaching the top of the academic distribution. In this sense, economic development and gender norms seem to matter. But the pipeline is only half the story.
A remarkably stable choice gap across all ten countries
Despite the wide variation in academic preparation, the choice gap – what high-achieving women and men want to study – looks strikingly similar everywhere. In every country we study, high-achieving women are about 20 percentage points less likely than men to choose STEM subjects.
This pattern emerges across the entire sample. In Sweden, one of the most gender-equal countries in the world, the choice gap is roughly the same size as in BrazilandUganda. In countries where women dominate the top of the academic distribution – Finland, Greece, Spain, Australia – the choice gap persists.
Even where women are a minority among top scorers – Uganda, Chile, Taiwan– the gap looks nearly identical. This stability is striking. These countries differ in income levels, education systems, labour markets, gender norms and cultural expectations. Yet high-achieving young women everywhere are consistently less likely to prefer STEM.
Why is the choice gap so stable?
Our data does not allow us to identify mechanisms directly, but a large body of evidence points to several possibilities. There may be information differences, where women might be receiving less accurate information about STEM careers, job prospects and earnings. Stereotypes and role models may also play a role, creating persistent beliefs about who “belongs” in STEM that shape confidence and identity.
There are also anticipated workplace norms: concerns about discrimination, harassment or demanding work cultures may deter women. Finally, preferences for job characteristics could result in STEM jobs being viewed as less flexible or more competitive, which may matter for students placing greater weight on their work-life balance. These forces are not unique to specific countries. Rather, they appear to operate globally, which may explain why the choice gap is so consistent across such different settings.
Tackling both the pipeline and choice gaps
Our findings underscore an important point: improving girls’ academic achievement will not, on its own, close the STEM gender gap. Sweden shows why. The country has essentially reversed the academic gender gap, and women now dominate the pipeline. But the choice gap is nearly identical to that of countries with lower gender parity.
This means that interventions aimed only at performance will leave the overall STEM gap largely untouched. Instead, policies must also address the choice side. This means providing better guidance and information about STEM fields, as well as more exposure to female STEM professionals.
Programmes that challenge stereotypes and build confidence are needed, as well as improving workplace flexibility and safety while showcasing the diversity of STEM careers beyond narrow images. Closing the STEM gender gap means addressing both access and preferences – not just one or the other.
This article is based on CEP discussion paper No 2120 Pipeline vs. choice: the global gender gap in STEM applications.
Note: This article gives the views of the authors, not the position of LSE European Politics or the London School of Economics.
Image credit: Matej Kastelic provided by Shutterstock.
Discussion about this post