So while it is randomized in the terms of who they chose to participate, the students did attend already existing schools. This could lead to selection bias from the participates, as the schools themselves are located in wealthier areas as that is where their clientèle are.

Lower income families may not have been able to take advantage of the lottery due to distance constrains thus self-opting out.

I have not read the study methodology details, the schools may have been chosen to avoid this problem but just wanted to point out that just because something say "random lottery" it may not be.