Re-Analysing the Transmission of Gender Attitudes from Teachers to Students: A Computational and Robustness Reproduction

This replication study evaluates the computational and robustness reproducibility of the main findings in Mehmood et al. (2025), which investigates the transmission of gender-equitable attitudes from teachers to students in Pakistan by means of a randomized controlled trial with four treatment groups and a control group. Using the authors' replication package, we perform a successful computational reproduction of all reported coefficients and p-values, starting from the provided analysis data set. We then perform four sets of robustness checks: using an alternative way of constructing a key outcome variable, including different sets of control variables, reweighting the student-level regressions, and applying a difference-in-differences approach based on pre-treatment outcomes. Overall, the main findings are highly robust to a wide range of reasonable alternative specifications and analytic choices. In particular, across the 90 robustness specifications we estimate, 94% reproduce the original sign and statistical significance of the reported effects. For some outcomes, we obtain considerably larger effect sizes, further underscoring the importance of the transmission of gender-equitable attitudes.