High School Students Drop Science when Scientists Become Role Models

At the end of high school, there is one policy question that appears simple and direct: how can schools broaden students’ career horizons when they need to reach thousands of teenagers, and when those teenagers have very different interests? A large online experiment published in Nature Human Behaviour could give the answer. Role models can change what students actually choose in college. The counter-intuitive part is the direction. When students saw videos of both successful scientists and entrepreneurs, they moved away from science, technology, engineering and mathematics majors. In the study, "role models" were young Ecuadorian scientists and entrepreneurs who described their careers, motivations, challenges and advice through a series of short online interviews.

We recruited young scientists and entrepreneurs from across Ecuador and asked them to speak candidly about their work, what drew them into it, the skills it requires, and the advice they would give to teenagers considering a similar path.

The finding of the nine-year study travels beyond Ecuador as other developing nations struggle with human capital , education finance and scaled learning technology. Indeed, to our knowledge this is the first time this has been observed. 

A national test built into school policy

The experiment was part of Showing Life Opportunities, a larger collaboration with Ecuador’s Ministry of Education, designed to promote learning in science and entrepreneurship. It was inserted into remote schooling during the pandemic for final-year technical high school students. Including a pre-test, the wider research program touched close to 35,000 students.

The core trial involved 29,243 students in 813 high schools. Half the schools were assigned to the role model videos; the others watched regular educational content from the Ministry’s EducaTV channel, on topics such as love and music. Students in the treatment group watched two 30-minute lessons made from interview segments with ten Ecuadorian scientists and ten Ecuadorian entrepreneurs. Each student saw ten scientist segments and ten entrepreneur segments.

The videos were designed to offer more than a single heroic story. The scientists included an electrical engineer who had developed a device to help people with limited vision, an environmental scientist with a patent using crab waste for environmental remediation, and a molecular biologist working on cancer. The entrepreneurs included business owners in sustainable coffee, tourism and creative baby clothing. The role models were young Ecuadorians, selected across gender, region and ethnicity. At scale, 94% of treated students watched between 19 and 20 planned segments. The calculated cost was $0.86 per student. As the research team later framed it, the goal was to ask what can be done at high school level, where numbers are large and students’ career interests are diverse.

The result moved against the aim

The intervention was meant to inform and inspire students to consider STEM careers, or high-growth entrepreneurship. But the results told another story. Among girls who entered college, assignment to the videos reduced enrollment in those majors by 5.0 percentage points, an 18% reduction relative to the control mean, and increased business enrollment by 5.7 percentage points. Business majors were already female dominated, and the treatment made them more so.

Among boys entering college, the videos reduced enrollment in STEM majors by 4.9 percentage points, a 7% reduction relative to the control mean. Boys shifted mainly toward other fields, with the study identifying increases in agriculture, humanities and arts from low starting levels.

The Ecuador setting also shows why this question matters in other developing-country school systems. In the sample, one third of students enrolled in higher education. Among control-group students who did enter college, 68.6% of boys chose science, technology, engineering and mathematics majors, compared with 28.5% of girls. 

What students appeared to compare

The students saw successful people in two broad fields during the same lessons, which gave them room to compare. A later scoring exercise asked 110 students to watch the videos and rate the role models on seven characteristics: passion, charisma, similarity, reinforcement of gender stereotypes, how challenging the subject appeared, how much relevant information the video provided, and stigma.

Those dimensions help explain the direction of the effect. Students saw the scientist role models as making their fields appear more challenging than entrepreneurship. They also viewed the science-related path as more male dominated and more stigmatized, and rated the scientist role models lower on relevance, passion, charisma and similarity to themselves.

We believe that the contrast between fields appears to have shifted students away from the path they saw as more challenging. The results also suggest that seeing many role models can lead students to compare commonalities and differences across fields, rather than attach themselves to one especially compelling individual story. -Compared to all previous role mode studies that involved meeting a single role model, having the opportunity to see several different role models probably led students to compare the commonalities and differences across them, eventually leading some students to be more likely to shy away from STEM.

Gender balance changed little

The study also tested whether the gender mix of role models mattered. On average, students saw a balanced mix, while random assignment meant some saw more women or more men in a category. Major choices varied little with that gender mix. Seeing mostly female scientist role models did not meaningfully offset the main effect.

For policymakers, the study’s impact lies in two findings that sit together. Digital role models can be delivered cheaply at national scale, with high completion, and can change real enrollment behavior. At the same time, giving students more than one field to observe can move choices in directions policymakers did not expect.

Providing a wider scope

This multi-field approach appears to be less paternalistic than trying to nudge students into one career path. It gives young people broader information. It also places careers side by side. In this case, the side-by-side comparison made STEM appear harder, more male dominated and more socially stigmatized than entrepreneurship.

Our paper stops short of saying whether the change made students better or worse off. That would require following them for years to measure earnings, satisfaction and labor-market outcomes. We can affirm that what is best for an individual student may differ from social priorities when scientific innovation creates wider benefits. The finding, then, is practical: exposure works, scale is possible, and the content mix shapes the direction of choice.