Applying the Regression Tree Method to Predict Students’ Science Achievement

Abstract

Science teaching is one of the most important tools for developing individuals’ critical thinking skills. Therefore, studies about achievement predictors of science teaching are increasingly being performed in order to provide evidence of what could influence students’ academic performance. As a contribution to those studies, this paper applies a predictive analysis to students’ science achievement in the 2011 National Examination for Secondary Education (ENEM). The sample is composed of Brazilian students who took the 2-day exam ENEM in 2011. The CART algorithm was applied through a model with 53 predictors. The model explained 24.50% of the science achievement variance. The results showed a lower achievement for those students who (1) were not enrolled in a private school to attend Secondary Education; (2) are female; (3) live in the North, North East, and Center-West regions of Brazil; (4) were strongly motivated to take the exam to obtain a Secondary Education certificate or scholarship; (5) had not yet finished Secondary Education until 2011; and (6) whose family income was equal or lower than a 1.5 minimum wage, as well as equal or lower than 5 minimum wages, depending on the type of school attended by the student in Secondary Education.