Efeito-escola e fatores associados ao progresso acadêmico dos alunos entre o início da 5a. série e o fim da 6a. série do ensino fundamental: um estudo longitudinal em Escolas Públicas no Município de Belo Horizonte - MG

Abstract

The research on schools effects is new in Brazil. Its first results have shown large educational nequalities and, in agreement with the previous international literature, that the out of school actors explain most of the variation on students proficiencies. Most of this research is based on ata collected by SAEB Brazilian System of Basic Education Evaluation. Although these esults are very relevant, they depend on cross sectional data, while it is well know that the most ppropriate form of measuring the school effects is through longitudinal data. Only this kind of ata allows the examination of the schools internal policies and practices on the student earning. This research uses a longitudinal, quasi-experimental design and its main objective is to measure the effects of seven public schools located in the same region of the city of Belo Horizonte, state capital of Minas Gerais Brazil. All these schools serve relatively homogenous ommunities in socioeconomic terms. All fifth grade students, in the chosen schools, were followed and tested three times, i.e., at the eginning of the first year, at the end of the first an second years of follow up. In every ccasion they answered a Portuguese language and a mathematics test. Some of the items used n the tests were taken SAEB, and therefore the students proficiencies could be expressed in the SAEB national scale. Besides the tests, the students filled a questionnaire with items on their emographic, socioeconomic and cultural characteristics. Additional items in the questionnaire aptured the students studying habits and their academic trajectory. Some parents and school taff were also interviewed in order to help the contextualization of the results, obtained with the chievement tests and questionnaires. The study described the student trajectory and explored the personal and school factors that nfluence students reading and math achievement and learning growth. The multilevel models itted to the data provide estimates the student-level variables that affect achievement and earning growth. Due to their reduced number, the schools are not a level of analysis in the multilevel models. Instead, they are included as dummy variables and the coefficient of these ariables in the models provide the schools effects estimates. The descriptive results reveal that students who had low initial status tend to growth more in the eriod. The multilevel models show that the student background factors (SES, sex, age gap) ave much impact in the status initial, but only some have impact in their rate of growth, epending on it disciplines or the grade. The qualitative evidences show that all schools analyzed have infrastructure problems and ttend students with modest socio-cultural backgrounds. These two characteristics have rofound impact on the students proficiencies. However school policies can either moderate or xacerbate the influence of these factors on the student achievement. Also the parents nterviews identified several family practices associate with school stratification. Together hese facts reinforce the necessity of public policies that would provide better schools to tudents from families with little socioeconomic resources improving the chance that public chools could compensate for the inequalities in the student origins. This study also provides important methodological contributions to the research on school ffects in Brazil, with longitudinal design. It also innovates on methods to treat concomitantly uantitative and qualitative information, again in the Brazilian scene.