The central claim of school choice proponents is that increased choice will give poor and minority kids access to higher quality schools and thus close the achievement gap. But we actually have very shaky evidence about the effects of schools of choice because of students' non-random assignment to these schools.
Along comes a spectacular new paper by economists Julie Cullen and Brian Jacob on school choice in Chicago, which provides the most rigorous evidence to date on this question. Chicago has one of the largest school choice systems in the country - more than a third of Chicago elementary students attend a school other than the neighborhood school assigned to them. Cullen and Jacob exploit the fact that most schools of choice assign slots by lottery in kindergarten and first grade; with longitudinal data, they can follow these students for five years.
What do they find? While lottery winners attend schools that are of higher quality in terms of their peers' achievement and the value-added by the school, it turns out that attending a school of choice confers no academic benefits on students, and these effects are consistent across demographic subgroups. Their conclusion has significant implications for current education policy, including the choice provisions of NCLB and many urban districts' efforts to expand choice:
Given the multiple disadvantages faced by poor families and the multiplicity of support services, along with the uncertainty regarding the impact of school quality on student outcomes, simply attending a better school may not be the most effective intervention....The original analysis conducted in this chapter suggests that schools are a blunt instrument for improving the achievement of disadvantaged students....We cautiously conclude that access to "better" schools is likely to be less effective than more targeted interventions.
See also the Cullen, Jacob, and Levitt paper on school choice in Chicago at the high school level, which also finds no academic benefits but important social benefits (i.e. fewer disciplinary problems and lower arrest rates).