The best article in this month's Smithsonian magazine is the profile of
Caroline Hoxby, who has studied the relationship between family income and the likelihood of a high-performing student applying to a top tier school. Maybe the most interesting tidbit of information is the relationship between income and students who perform at the 90th percentile or better on the college tests.

Yes, it's weighted toward richer students, but honestly it is nowhere near as skewed as I expected. The poorest 25% of families still produce 17% of these higher achieving students. BUT, these students just do not apply to Harvard. Harvard wanted to do something about it, so they made tuition FREE for poorer students. This had very little effect on their applications pool. So Hoxby's next step has been to work with the ETS so that high performing students on the college tests get a special packet, which not only encourages the students to apply to top schools, but also includes vouchers for free applications to these schools.
The other interesting tidbit was that, although many schools focus on urban areas to locate disadvantaged students of promise, her work discovered that there is also a lot of untapped potential in smaller towns and rural areas, where bright students do not choose to apply to faraway top tier schools.