Thanks for the approach, I'll need to read about those models and how they correlate with real data. It seems likely that the gender gap appears from a combination of incentives and prejudices, so ideally we should use a combined model that describes how much each factor affects the final outcome.