Brazilian national assessment data and educational policy: an empirical illustration. Issue 1 (2nd January 2016)

• Record Type: Journal Article
• Authors: Paget, Christine L.; Malmberg, Lars-Erik; Martelli, Dale R.
• Is Part Of: Assessment in education. Volume 23:Issue 1(2016)
• Page Start: 98
• Page End: 125
• Publication Date: 2016-01-02
• Subjects: national assessment systems -- Brazil -- school effectiveness -- school resources -- educational policy evaluation; Educational tests and measurements -- Periodicals; Education -- Evaluation -- Periodicals; 371.26
• DOI: 10.1080/0969594X.2015.1113929
• Languages: English
• ISSNs: 0969-594X

Abstract

Abstract : In concert with other Latin American countries, Brazil has developed and implemented its own national assessment system for the purpose of monitoring, evaluating and improving their educational system. Prova Brasil is a census-based bi-annual assessment of Portuguese and mathematics achievement of middle school students in Brazil accompanied by four background surveys of students, teachers, principals and schools. This study uses the Prova Brasil assessment data to evaluate the Brazilian educational policy objectives of improved educational achievement through increased school resources using the North-Eastern state of Paraiba, one of the country's poorest regions, as a sample base. We analysed a sample of 166, 354 students in 740 schools over three time points (2007, 2009 and 2011) and two grade levels (grades 4 and 8) separately. Predictor variables were entered hierarchically into two level multilevel models. The study found that infrastructure and academic resources significantly predicted Portuguese and mathematics achievement. This study contributes to the growing body of evidence that school resources have an impact on educational outcomes in the developing world, in contrast to the lack of evidence for impact in developed countries. The study is limited in a number of ways and therefore results should be interpreted with caution primarily due to the cross-sectional nature of the data, the voluntary and low-stakes nature of the testing, and the fluctuating student sample sizes.