Predictive Model for School Dropout in Chimborazo Province, Ecuador

Abstract

Introduction: School dropout is a complex problem influenced by various factors, including disparities in educational quality, inadequate infrastructure, and adverse socio-cultural conditions. This phenomenon negatively impacts the social and economic development of the country. Despite the recent decrease in dropout rates in Ecuador, the problem remains significant. Objective: To develop predictive models, including linear regression and generalized linear models in R-studio, to forecast dropout rates and identify significant institutional and demographic factors. Method: A quantitative approach was adopted to analyze data from the Ecuadorian Ministry of Education for the periods 2009-2010 to 2023-2024. Data on enrollments, approvals, non-approvals, and dropouts were reviewed using descriptive statistics and correlation analysis. Results: The results showed a decrease in dropout rates starting from the 2013-2014 academic year, although with significant fluctuations. Higher dropout rates were identified in public institutions and rural areas in the Sierra region, specifically in public institutions in Chimborazo province, accounting for 97.47% of the total dropouts, in contrast to students from the Coastal región. Additionally, a p-value of 0.073 was obtained in the linear models, so the null hypothesis was not rejected, suggesting that the residuals are approximately normal. Conclusions: The predictive models (LM and GLM) effectively estimated dropout rates in Chimborazo, with the GLM showing a slightly better fit. The type of institution and geographic location were significantly associated with dropout rates, highlighting the need for interventions targeting public institutions and rural areas. Strategies to reduce dropout rates should focus on improving conditions in these specific areas