Gender and academic indicators in first-year engineering dropout: a multi-model approach

Student attrition during the first academic year remains a critical issue in engineering education, with implications for equity and institutional effectiveness. This study explores early dropout patterns across five engineering degree programs using data from 3889 first-year students spanning seven academic cohorts at one public and one private university in Spain. A multi-method analytical strategy—comprising logistic regression, artificial neural networks, and propensity score matching—was used to examine how gender and academic performance relate to early attrition. Findings challenge prevailing assumptions in the literature by showing that gender is not a statistically significant factor in first-year dropout, a result consistent across methods and after adjusting for confounders. In contrast, academic variables—such as entrance exam scores, failure rates, and exam absenteeism—exhibited strong associations with attrition. Institutional context also shaped dropout patterns: academic failure played a particularly salient role in the private university, while disengagement, measured through no-show rates, was more relevant in the public institution. Notably, an exception emerged in the Mathematical Engineering program, where gender moderated the link between academic failure and dropout, pointing to potential interaction effects in specific curricular settings. These results underscore the importance of understanding the conditional and context-dependent nature of early dropout, supporting targeted interventions grounded in academic, rather than demographic, indicators.