Performance of public-school students in the National High School Exam (2012-2018): analysis and representation

Abstract

Data was analyzed from 5.6 million public school students who scored in all tests of the National High School Exam (ENEM), from 2012 to 2018. Student performance was grouped by school and by zone and correlated with socioeconomic and school variables. Student performance was also represented in different administrative units (municipality, micro and mesoregions, states and regions). The modifiable areal unit problem and fallacies were also evaluated in the comparison in different levels of aggregation. The analysis determined that the average score of graduates improved in the time period, but the total number of students with high proficiency did not increase in the same way. Of the variables analyzed, family income per capita showed a greater correlation with performance in the educational tests. An analysis was also conducted for the problem of modifiable areal unit problem that results from the change in average results and correlations in the aggregation of Federation Units in Regions. The variation of means and correlations in the different administrative zonings of the Brazilian territory was analyzed as well as a determination of some of the possible fallacies that could occur in the interpretation of the results. Although the amount of data analyzed was substantial and results of the analysis have allowed identification of the areas of best performance, it was determined that the low percentage of participation in the exam compromises the potential of using this test to assess the educational quality of Brazilian public schools. In most cases, the lowest proficiencies in Brazil occurred in the municipalities and micro-regions of the North and Northeast of Brazil, where the average per capita income is often lower. Although correlation values increased with a greater aggregation level, the perception of this association between variables on maps decreased.