This paper aims to measure the difference of performance between state and private primary schools. We use weighted least squares by propensity score method along with SAEB-2003 data set to avoid selection problems and to have better causality estimation. The results show a huge difference of performance between these schools students and an increasing difference according to student's socioeconomic level and the average salary paid for teachers. Black and mulatto students get less benefit with private schooling than white and Asian ones. Increasing spending do not reduce such differences, these also vary according to non observed regional differences.
microeconometric; education; propensity score