A relação entre a expansão do acesso ao ensino e o desempenho escolar no Brasil: evidências com base no SAEB para o período de 1997 a 2005

Abstract

The success achieved by the policies to make educational opportunities more democratic and the increase in vacancies for elementary school brought about new challenges to the Brazilian educational system. In the last years, the problem related with the poor quality of education, measured by the students performance in standardized knowledge tests have been widely discussed in the literature.Since 1995, information on the performance has been made available at Sistema de Avaliação da Educação Básica (Primary Education Evaluation System - SAEB). When the tendency of measures is assessed, it is possible to see a decrease in proficiency levels by Brazilian students. The first assumption determined to explain this negative evolution is based on the increase in students heterogeneity due to the policies to provide more access to education and to reduce the number of students that fell behind in the school, introduced in the second half of the 90s.The present study uses the quality and resourcefulness of historical data on students performance to explore the variations of this measure and to find possible explanations for these fluctuations. There have been many studies on the factors associated with school performance at a certain point in time, however, little attention has been paid to the changes in these factors and their effects on the intertemporal variations observed in school proficiency. To find and quantify these effects was the main objective of the present study which, to that end, applied decomposition methods based on counterfactual simulations.In the first part, univariate analysis has been performed to assess the differences in school performance distribution between the two periods. We have used relative distribution methods developed by Handcock and Morris (1999), and it was seen that temporal decline in average school performance was because of the connection of two factors: a decrease in the performance levels over distribution quantiles (level effect) and an increase in the relative participation of students with poorer education results (distribution effect).In the second part, school performance was conditioned to its associated factors and we have tried to identify the contribution of changes in the composition and return of heterogeneity observed and unobserved in the temporal variation of the average and of the distribution of school performance. In this exercise, we have used Oaxaca-Blinder (1973) and Juhn, Murphy, Pierce (1993) decomposition. Due to the hierarchical nature of educational data, we have adopted strategies that enabled measuring these effects in both levels: student and school. Among the main results found, we have seen that the effects associated with variation in students socioeconomic status were more important to explain the temporal variations in school performance, both in the micro analysis (student) and in the macro analysis (school). Reduction in the amount of this attribute (composition effect) contributed to reduce the average and to increase differences in students results in proficiency tests. On the other hand, the reduction in the sensitivity of school performance to socioeconomic status (return effect), although it has contributed to decrease average quality of education, it helped equalize the distribution of school performance in all periods assessed. As for unobserved heterogeneity (residual effect), we have seen that its contribution to the variation in inequality was small compared to the others.