Book Description

This research addresses gender peer effects in education and their impact on student achievement in Chile. We address the topic from three different level of analysis: (a) whether the proportion of girls in a cohort influences students' educational outcomes (b) whether assignment to a classroom with a higher proportion of girls influences students' educational outcomes, and (c) the relative performance of single-sex and coeducation in Chile and its impact on male and female achievement. Through three essays, peer effects are analyzed and their role established for different stages of schooling and for various student characteristics, including gender, socioeconomic background and type of school the student attends. Moreover, the possibility of nonlinearity in gender peer effects is explored as well at the various possible channels through which the effects may operate, such as changes in the amount of curriculum covered. The three essays also address the econometric problems inherent in any study of peer effects. The dissertation provides new evidence on the existence of gender peer effects in elementary and secondary schools based on data from a developing country. It also provides an enriched understanding of how gender peer effects are interrelated with characteristics of the students and schools. The analyses were based on recent data from SIMCE, a Chilean national standardized student assessment carried out in 4 th , 8th and 10 th grade. It is concluded that, after controlling for differences in socioeconomic background, school and a cohort's characteristics, a larger share of female students in a cohort and in a classroom level have a positive impact on academic achievement, both for boys and girls. Results were robust across estimations. Classroom gender peer effects were bigger than the ones estimated at the cohort level, supporting the notion that peer effects get larger when they are measured closer to the context in which they operate. The effects are slightly stronger for girls. The estimated impact of an increase in 10 percentage points on the proportion of female students in the classroom is about 2 to 8 percent of one standard deviation in girl test scores, and between 1 to 5 percent for boys. Gender peer effects proved to be nonlinear and they were stronger when female students are in a minority within the student population, both for boys and girls. To explain these results, the research shows that, as the share of female students in a cohort or class rises, the amount of curriculum that a teacher can cover during the academic year increases as well. Furthermore, the increased share of females is associated with an increase in teachers' expectation on academic attainment of their students. The analysis of co-education and single-sex schools concludes that most of the differences in student achievement between students that attend a single-sex versus a coeducational school were due to students' background characteristics, previous student achievement, peers and school selection. The issue of selection bias is much weaker for girls. Some positive and statistically significant positive effect of single-sex school attendance persist for girls after the effect of selection is controlled-for, as well as previous achievement and other controls. The difference is small, though. The single-sex school effect almost disappears for boys, supporting results from previous sections of this dissertation.