Palestrante: Reynaldo Fernandes (USP / Ribeirão Preto)
This paper proposes a school quality measure that is less subject to measurement errors and also brings more information about the influence the school has over the students’ achievement, relatively to those commonly used on school accountability programs. It is proposed a quality measure that is formed by a weighted average of other measures of school quality, including results, inputs and processes. The weight structure is defined by a MIMIC model (Multiple Indicators and Multiple Causes) and not of arbitrary forms. The results’ indicators used are value added measures, calculated by the fixed effect and between effect models. We applied this method for the municipal schools of the city of São Paulo using 2007 and 2009 Prova Brasil’s data. The results indicate that our quality measure is less stable than raw scores but the correlation with the socioeconomic characteristics of students is smaller. Not surprisingly, this measure also was highly correlated with the result indicators that compose it. In our next step we will work with indicators from other data sources (e.g. other years, grades and test).
Data: 13/11/2013 (quarta-feira)
Sala Paulo Renato de Souza – 2º andar