In Brazil, inequality of educational outcomes by parental education is well documented (Torche 2014, Portela 2013, Marteleto 2012, Alves et al 2013, Ribeiro 2011, Soares and Colares 2006), as well as the inequality in educational outcomes among Brazilian whites, blacks and mestizos (Silva 2000, Fernandes 2004, Botelho et al 2010). Yet, the literature is limited to studies based on cross-sectional data presenting snapshots of social inequalities at single or few points in time. The trends over time in achievement gaps and segregation by social background characteristics, such as race and parental education, have never been addressed.
In this study, I calculate trends on achievement gaps and segregation by race and parental education for 5th, 9th and 12th grade students from 1995 to 2013, drawing on data from SAEB, the Brazilian National Assessment of reading and math performance of children in public and private schools (SAEB – Sistema de Avaliação da Educação Básica). I use the V-stastistic (Reardon and Ho, 2014) to estimate reading and math gaps between whites and blacks, whites and mestizos, students with parents in the top and bottom decile of the parental education distribution, students with parents with more and less than 8 years of education. The estimation of trends in segregation relies on Theil Information Theory Segregation Index (Reardon and Firebaugh 2002). I use Oaxaca decomposition models to analyze to which extent the changing social composition in schools affect the trends in achievement inequality by race and parental education.