This paper presents the methods used in a school effects study of seven public schools from the same geographical area of Belo Horizonte, state capital of Minas Gerais - Brazil. The school effects research, a sub-area of Sociology of the Education, deals with the schools impact in student achievement, after the individual student socioeconomic status and pertinent school contexts variables are accounted for. During the two years that the study reported here lasted, the students' achievement was measured through their results in three tests and family and school staff context data through several interviews. Item Response Theory (IRT) models were used to estimate and equalize the students' proficiency in the three waves of the study. The socioeconomic status measure was constructed using the similar IRT models. Regression models for hierarchical data were adjusted to analyze the longitudinal data nested in different levels of aggregation. Software that allows the integration of textual and numerical data helped the analysis of the data collected through the interviews. Longitudinal data allows the use of more appropriate measures of school effects since it captures the student learning and not only the student achievement. The qualitative data analyses expose the school and familial factors that affects the results of the pupils and contribute for the pertaining to school stratification. Considering the potential impact of educational research on policy, the paper emphasizes the importance of using adequate models in the analysis of longitudinal data for the estimation of the school effects. Moreover, this study shows the viability of the integrated use of quantitative and qualitative methodologies, which are usually treated as opposite paradigms, and the relevance of this procedure for the understanding of the general results.
Sociology of Education; school effects research; quantitative methods; qualitative methods
