Universality, Correlations, and Rankings in the Brazilian Universities National Admission Examinations

TL;DR

This study analyzes Brazilian university entrance exam scores, revealing high correlations across subjects, bimodal high school score distributions linked to socioeconomic factors, and differences between standardized and non-standardized test performances.

Contribution

It provides new insights into the distribution and correlation of exam scores, highlighting limitations of using high school scores for institutional evaluation.

Findings

High correlation between different subject scores in ENEM.

High school scores form a bimodal distribution influenced by socioeconomic status.

Standardized and non-standardized exams select different individuals.

Abstract

The scores obtained by students that have performed the ENEM exam, the Brazilian High School National Examination used to admit students at the Brazilian universities, is analyzed. The average high school's scores are compared between different disciplines through the Pearson correlation coefficient. The results show a very large correlation between the performance in the different subjects. Even thought the students' scores in the ENEM due to the standardization form a Gaussian, we show that the high schools' scores form a bimodal distribution that can not be used to evaluate and compare performance over time. We also show that this high schools distribution reflects the correlation between school performance and economic level of the students. The ENEM's scores are compared with a Brazilian non standardized exam, the entrance exam at the Universidade Federal do Rio Grande do Sul. The comparison of the performance of the same individuals in both tests is compared showing that the two tests not only select different abilities but chooses a different set of individuals. Our results indicates that standardized exams might be an interesting tool to compare performance over the years but only of individuals and not of institutions.