Resumo

HENRIQUEZ C., Natalia  e  ESCOBAR R., Diego. Construction of an Early-Alert Model for Detecting Students at Risk of Dropping Out of Educational Science in Universidad Metropolitana. RMIE [online]. 2016, vol.21, n.71, pp.1221-1248. ISSN 1405-6666.

This article analyzes the variables that are associated with the skills required upon enrolling in Educational Science at Universidad Metropolitana. The purpose is to establish indexes for detecting students who require support during the first academic year, which is characterized by high dropout rates. The modeling was done through logistical regression with ROC curve analysis. The results showed that the variables that allow predicting deficiencies in language and mathematical skills were: scores on the university admissions test in both subjects, average grades in secondary school, gender, and age upon enrollment in Universidad Metropolitana. Based on these variables, a student classification was developed (cut-off points and probabilities) to establish strategies of academic support.

Palavras-chave : dropping out; prevention; academic factors; statistical methods; Chile.