Estudo de fatores associados através de regressão quantílica hierárquica

In this paper, we present an unusual approach to factors associated with academic achievement since we use hierarchical quantile regressions. While the traditional approach aims to identify important factors based on individuals with an intermediate performance, our approach aims to detect the effects on performance distribution quantiles, thus allowing to identify how a certain factor can affect poor, intermediate-, and high-performing pupils. We describe the methodology and use it with data from Portuguese and Mathematics tests of the 8th grade of primary education in the Brazilian state of Para, in 2016.