Um modelo de escolha discreta para avaliação de desempenho em concursos públicos

Abstract

The main objective of this thesis was to investigate the factors determining success in an aptitude test for admission in a public bank in the Northeast of Brazil. The data set is based in a questionnaire for 232,771 Candidates which is composed of the result of the test and a number of questions about some attributes of the candidate. Attributes selected were personal monthly income; household monthly income; education level; type of school attended (public or private); marital status; occupation and age. A logit model was used to estimate a regression of success in the test against the attributes defined above. Results of the estimated model show that all the coefficients but marital status were significant and with the expected sign. The marginal effects inform that a high school level (a requirement to take the test) had the strongest and negative impact on the probability of success. Specification tests for functional form and unobserved heterogeneity show that the model was correctly specified.