Does "early bias" theory include the leaky pipe line model? Ie: University students that fail a exam will switch program at a higher rate if their gender is a minority. In polls, they are also more likely to question their decision in career choice and have more self doubt about their ability to finish the program. The data that I have seen on this is that it is gender neutral, ie that the doubt caused by being a local minority is similar for both men and women.

To make up numbers, if a student has 3 times a high risk of switching a program after a failed exam, the chance that they will graduates will be significant lower. Having a gender segregated education fixes this, but as soon the graduate enters the work market the same phenomenon happen after a failed interview unless they only seek work which is also gender segregated.