The "school effect" in some Latin American countries: Reanalyzing SERCE data

Abstract

The Latin American Laboratory for the Evaluation of the Quality of Education (LLECE) at Regional Bureau for Education in Latin America and the Caribbean of UNESCO has recently released the results of the Second Regional Comparative and Explanatory Study (SERCE) on the relationship between some school and extra-school factors and sixth grade primary student learning in Mathematics and Reading. For data analysis, the LLECE team used hierarchical linear models with two levels: student and school. The partial credit model technique (Rasch) was applied for the construction of socioeconomic and cultural status of students´ families. Math and reading tests were analyzed separately. The SERCE publication presents the estimates of gross, adjusted and net school-effects resulting from the application of that methodology. In this article, SERCE´s data is analyzed using bivariate multilevel models with three levels (country, school, student) and a different procedure for the construction of socioeconomic and cultural status of families. The results obtained differ markedly from those reported by the LLECE. The article draws some conclusions and discusses policy implications.