Mixture Structural Equation Models for Classifying University Student Dropout in Latin America

Abstract This research seeks to develop a model that allows consider the different forms of heterogeneity of international dropout data and also classify students who continue studying and those who drop out. Specifically, through the use of Mixture Structural Equation Models (MSEM), the study seeks to develop a model for classifying dropout and applying it to an international database. The aim is then to determine the classification accuracy degree of the proposed model. The development and application of the model showed that the student´s health, the interpersonal relationships, and class attendance positively influence college adaptation, and in turn college adaptation positively influences college satisfaction. Additionally, the developed model can correctly classify 55.45% of continuing students and 61.68% of students who abandon their careers. These results suggest that the use of MSEM for international databases, characterized by heterogeneity, allows more robust and generalizable studies of dropouts in higher education.