Gender segregation in the labour market may be partly explained by women’s reluctance to choose technical occupations, although the foundations for their career choices are laid much earlier, during education. Educational experts claim that girls are doing better in math and science if they are in single-sex classes. Explanations might be the dominating behaviour of boys in the classroom, unequal treatment by teachers, or the lack of role models and self-confidence of girls in male-dominated subjects. In this paper we try to identify the effect of gender-composition on the school choice of female students in Austria. We propose that female students are less likely to choose a female-dominated school type and are more likely to choose a technically-oriented or male-dominated school type at the age of 14 after spending the previous years of secondary education in classes with a higher share of female students. The Austrian education system is organized in a segregate way and relies on early tracking of students into different school types. After finishing the 8th grade either in a low track school or an academically preferable high track school, students choose between different types of schools: a pre-vocational school, a range of intermediate and higher vocational schools or a higher general school. Higher and intermediate vocational schools provide for example technical education, commercial education or education in domestic sciences and tourism. We study the 9th grade school choice of girls in low track schools in Linz, an Austrian city of about 189,000 inhabitants, using register data from 1988 to 2006. We estimate the causal effect of the share of female students in grade 5 on female school choice in grade 9. Gender composition in lower secondary schools is endogenous, since students may select schools depending on the share of females or correlated unobserved characteristics, like core subjects. Following Hoxby (2000) and Lavy and Schlosser (2007), we address this problem by using natural population variation in gender composition of adjacent cohorts within schools. Using binary and Ordered Logit models we estimate whether girls are more likely to choose male-dominated school types and less likely to choose traditionally female-dominated school types (like vocational schools for domestic sciences) if they were in grades with a higher share of female students. We include school-specific quadratic time trends to control for school trends in unobserved factors that may be correlated with the share of females in a certain grade.

We use register data, covering compulsory school students from Linz between 1988 and 2006. We estimate our model using Binary as well as Ordered Logit regressions. The identification strategy is based on natural variation of gender composition of adjacent cohorts within schools.

The estimations show sizeable effects of gender composition for female students. A higher fraction of girls increases the probability to choose the technical vocational school and decreases the probability to choose a female-dominated school for female students. Our paper contributes to the recent debate about coeducation versus single-sex schooling either in certain subjects or at the school level.