
Modeling of university dropout using Markov chains
Author(s) -
José Alejandro González Campos,
Cristian Manuel Carvajal-Muquillaza,
Juan Aspeé
Publication year - 2020
Publication title -
uniciencia
Language(s) - English
Resource type - Journals
eISSN - 2215-3470
pISSN - 1011-0275
DOI - 10.15359/ru.34-1.8
Subject(s) - dropout (neural networks) , markov chain , equity (law) , index (typography) , randomness , estimator , university system , higher education , statistics , econometrics , mathematics education , computer science , actuarial science , psychology , economics , mathematics , political science , economic growth , machine learning , world wide web , law
Access to higher education is only a first step in achieving equity in education; the following step is improving student retention, or lowering dropout rates, which is the same thing. The present study focused on the definition of an index as an estimator of the risk of individuals dropping out of a university using a Markov chain model, based on the randomness of the occurrence of dropping out. The suggested index was applied to a sample of 5,700 university students from the 2012-2015 annual cohorts of 8 university departments of a public regional university in Chile. The results indicate that the highest average probability of dropping out (slightly more than 39%) occurs in the first 2 semesters of university studies, and then decreases through time. This indicates the need for institutional retention policies that pay particular attention to the first year of university studies. Having this index also allows a formal estimation of changes or temporary variations in the risk, as well as quantifying the impact of interventions, not only for the case under study but for the entire higher education system.