Analysis of the level of achievement of the bachelor's degree exam as a function of sociodemographic variables

Abstract

Introduction. The present work describes the multinomial logistic regression model with application to real data, using the database of the Ser bachiller exam - Ecuador of the 2018-2019 school year, modeling the level of achievement reached in the grade exam grade, based on the sociodemographic variables age, sex, province, type of financing of the educational institution and ethnic self-identification. Goal. Describe the use and usefulness of nominal response models and ordinal response models, obtaining their best predictors and establishing their differences in data modeling. Methodology. For the calculation, the free software R was used with data taken from the National Institute of Educational Evaluation, adjusting the nominal and ordinal response model, in addition, applying the stepwise method, the best predictors for the study were found. Results. The results indicate a positive and significant correlation between the control environment and the ROE and ROA indicators. Conclusions. It was possible to conclude through the application of the stepwise method that the independent variables used in the study are the appropriate ones, since through the automatic selection in both the nominal and ordinal response models they give the same variables as predictors.