Logistic regression and prediction of low academic performance of students majoring in medicine

Abstract

Background: prediction of the academic performance of students can be useful for the institutions to devise strategies aimed at improving the development of the teaching-educational process.Objective: to determine the predictive capacity of some factors about the poor academic performance of medical students of the Affiliated School of Medical Sciences of Puerto Padre, University of Medical Sciences of Las Tunas, Cuba.Methods: an analytical, predictive study was carried out with 87 medical students enrolled at the aforementioned institution. The study was divided into two stages: in the first stage, by means of a bivariate analysis, the variables that were mostly associated with low academic performance were identified; and in the second stage, by means of logistic regression (multivariate analysis), the capacity of these variables to predict it was analyzed.Results: in the bivariate analysis, eight variables showed a significant relationship with low performance, when subjected to multivariate analysis (correlation and logistic regression), three of them maintained statistical significance, so they were finally chosen as predictive variables. According to this, the low academic performance can be predicted by the synergic combination of the predictors: school repetition, dedication of less than 15 hours per week to study and average grade in the university entrance examinations.Conclusions: a predictive model was determined, proving to be appropriate to assess the low academic performance of the students majoring in medicine.

Keywords

• academic performance
• education, higher
• forecasting
• logistic models