Student attrition in engineering programs remains a significant challenge for higher education institutions, with gender differences often considered a contributing factor. However, there is no clear consensus on whether gender directly influences dropout rates. This study examines the role of gender in the attrition of first-year Chemical Engineering students at a public university and how its impact evolves throughout the first academic year.

The analysis focuses on two critical stages: university entry and the completion of the first year. Logistic regression, propensity score matching (PSM), and a feedforward neural network with post-hoc analysis were applied to isolate the effect of gender from other confounding factors such as academic performance and no-show rate. The PSM method was particularly useful, as this technique allows inferring causal relationships in already observed data. The results indicate that gender is not a significant predictor of dropout at either stage, contrary to what has been reported in some previous research. Instead, failure rates and, more prominently, no-show rate are the strongest predictors. While lower university entrance exam scores correlate with dropout at the university entry, their impact decreases as students adapt to the academic environment.

Keywords: Attrition, Engineering, Gender influence, Dropout, Propensity Score Matching, Neural Network, Post-hoc Analysis.