In the studies on the effects of schools there is a clear distinction between pupil's performance at school and learning. The former notion refers to the level of performance at some stage of the process of schooling, whereas the latter is related to the acquisition of knowledge along the school trajectory. There is a consensus in the literature around the fact that, in order to study the success of individual schools, one has to analyze the latter notion using longitudinal data. The objective of the present study is to measure the effect of seven public schools of the city of Belo Horizonte, which are similar in terms of their location and socioeconomic profile. The study is based on longitudinal data collected during two years starting at the 5th grade. The students were subjected to cognitive tests at three moments. They also responded to an in-context questionnaire to gather information on their demographic and socioeconomic characteristics, amongst other data. The effect of schools on learning was analyzed with multilevel regression models for longitudinal data, in which the schools entered with indicator variables. The results have shown that, on average, students with lower initial level improved more than students with higher initial level, although differences in level remain at the end of studies. The socioeconomic attributes of pupils have large impact on the explanation of differences of initial level, but only some of them have impact on pupils' learning, depending on the discipline and grade. This study offers important methodological contributions to the investigation of the effect of schools under a longitudinal perspective.
School effect; Longitudinal research; Fundamental education
