Success Academy, a network of charter schools in New York, seems to be doing very well for its students. And their success seems to be the result of more than just selection bias [see this].
What if those Success Academy kids had been distributed among all of the top schools in the city? Would they have done better or worse in math?
Brown found that they would have performed significantly worse: 310 fewer kids would have gotten a level 4 on the exam, the top score. And 203 more kids would have received a 1 or a 2, which are considered to be failing grades. That would have been a tremendous waste of potential.
Of course, this assumes that all of these schools are drawing from the same pool. That's difficult to determine with any certainty, but we know that Success Academy students, who are selected through a random lottery, tend to come from disadvantaged backgrounds. [emphasis added] Seventy-seven percent of its students qualify for free or reduced school lunch. Ninety-two percent of test takers were black or Hispanic.
Let's hear it for private alternatives!