Academic performance and single-sex schooling: Evidence from a natural experiment in Switzerland

Female students in all-female classes experience less stereotype threat and perform better in their mathematics grades than their female peers in coeducational classes, with no difference in their language grades.


The gender gap in academic performance might be explained in part by stereotype threat, or the anxiety or concern that individuals of a certain identity (e.g., woman) feel when they risk confirming negative stereotypes about that identity (e.g., women’s inferior mathematics ability). Here, it is suggested that having any male students in the classroom might prime gender-based stereotypes for female students, leading to worse performance on stereotypically “male” subjects such as mathematics. However, it is hypothesized that there would be no difference in performance in non-stereotypically “male” subjects such as language.

It has been suggested that single-sex classrooms in these stereotypically “male” subjects may be advantageous for female students. If single-sex schooling reduces gender-specific stereotype threat, it is hypothesized that girls in single-sex classes would do better in their mathematics grades than girls in coeducational classes. The researchers take advantage of a natural experiment due to a Swiss high school administration’s random assignment of female students to either all-female or coeducational classes to study the peer effects of classroom gender composition on academic performance of female high school students. The students’ academic performance is observed over a time period of up to four years.


Single-sex schooling improved the performance of female students in mathematics. This positive effect was particularly large for female students with high academic achievement prior to entering high school. Single-sex schooling also strengthened female students’ self-confidence in their mathematical abilities and increased their chances of attributing these abilities to their own efforts rather than outside factors such as talent or luck.

  • In mathematics classes, single-sex classes had positive benefits for female students:
    • Female students in single-sex classes obtained better math grades than female students in coeducational classes, with an average performance increase of approximately 7–10% within the range where most students score.
    • Single-sex schooling benefitted female students regardless of prior abilities. However, female students with higher estimated academic abilities benefitted more than less talented female peers with a one point higher score on the math qualifying exam associated with 45% of a grade point more in the single-sex classroom and only 31% of a grade point more in the coeducational classroom.
    • Single-sex schooling benefitted female students regardless of teacher gender. However, the effect was smaller for girls in classes with female teachers compared to male teachers.
    • In addition, female students in single-sex classes evaluated their mathematics skills more positively (average 2.382 vs. 2.032 on a scale of 1-4) and were more likely to attribute their performance in mathematics to their own efforts rather than to talent or luck (average 3.402 vs. 3.051 on a scale of 1-4).
  • In language classes, single-sex classes had no effects on female students:
    • Female students in single-sex classes performed equally to female students in coeducational classes.
    • Teacher gender had no impact on female students’ performance in either single-sex or in coeducational classes.
    • There was no difference in how female students viewed their language skills in single-sex or coeducational classes.

In sum, single-sex schooling improves the performance of female students in mathematics classes but not in language classes, suggesting that reducing gender-based stereotype threat has real effects on academic performance. Moreover, female students who demonstrated high pre-existing ability on the entrance exam benefitted the most from single-sex classrooms, which underlines the relationship between the beneficial effects of all-girls schooling to the absence of gender-specific stereotype threats.


The researchers exploited a natural experiment at a high school in the German-speaking part of Switzerland. Like most Swiss high schools, this school is run and financed by the local canton (state) and applies standard curricula and teacher recruitment policies. Since the school has a focus on teaching pedagogical subjects, which allows many of the school’s graduates to skip the first-year courses at the University of Teacher Education (located in the same town), it attracts many female students, and 80% of students are female. In order to provide male students with more peers of their gender in classes, the school board arbitrarily assigns incoming female students to coeducational or single-sex classes for the entire four years of education. Students and their parents cannot influence the assignment, so the assignment is de facto random. The coeducational classes have about 22% male students. All students experience the same curriculum and examinations in the core subjects, taught by the same teachers, which means the students’ grades in math and German are comparable across parallel classes.

The researchers observed the performance of 808 high school students (668 female) who entered the school (9th to 12th grade) between 2001 and 2008 in both math and German, which are compulsory courses. The researchers also conducted a survey to gauge female students’ experiences in their math classes and beliefs about their mathematic abilities.

Related GAP Studies