Early Dropout Predictors in Social Sciences and Management Degree students
Student dropout is a major concern in studies investigating retentionstrategies in higher education. This study identifies which variables areimportant to predict student dropout, using academic data from 3583first-year students on the Business Administration (BA) degree at theUniversity of Barcelon...
| Autores: | , , , |
|---|---|
| Tipo de recurso: | artículo |
| Estado: | Versión aceptada para publicación |
| Fecha de publicación: | 2024 |
| País: | España |
| Institución: | Varias* (Consorci de Biblioteques Universitáries de Catalunya, Centre de Serveis Científics i Acadèmics de Catalunya) |
| Repositorio: | Recercat. Dipósit de la Recerca de Catalunya |
| OAI Identifier: | oai:recercat.cat:2445/215840 |
| Acceso en línea: | https://hdl.handle.net/2445/215840 |
| Access Level: | acceso abierto |
| Palabra clave: | Abandó dels estudis (Educació superior) Rendiment acadèmic College dropouts Academic achievement |
| Sumario: | Student dropout is a major concern in studies investigating retentionstrategies in higher education. This study identifies which variables areimportant to predict student dropout, using academic data from 3583first-year students on the Business Administration (BA) degree at theUniversity of Barcelona (Spain). The results indicate that two variables,the percentage of subjects failed and not attended in the first semester,demonstrate significant predictive power. This has been corroboratedwith an additional sample of 10,784 students from three-degreeprograms (Law, BA, and Economics) at the Complutense University ofMadrid (Spain), to assess the robustness of the results. Three differentalgorithms have also been utilized: neural networks, random forest, andlogit. In the specific case of neural networks, the NeuralSensmethodology has been employed, which is based on the use ofsensitivities, allowing for its interpretation. The outcomes are highlyconsistent in all cases: both a simple model (logit) and moresophisticated ones (neural networks and random forest) exhibit highaccuracy (correctly predicted values) and sensitivity (correctly predicteddropouts). In test set average values of 77% and 69% have beenrespectively achieved. In this regard, a noteworthy point is that onlyacademic data from the university itself was used to develop themodels. This ensures that there’s no dependence on other personal ororganizational variables, which can often be difficult to access. |
|---|