The message of Ready et al.'s analysis is clear: The seeds of academic success or failure are sown long before high school. Since variations in students' academic performance are on average much larger within a given school than across schools, which school a student is enrolled in is far less important than a student's experiences in whatever school he or she attends. Rather, the similarity of trajectories of success or failure in different schools suggest that early diagnosis and intervention are likely to be more successful strategies than tinkering with the system of school choice, especially at the elementary and middle school level.

The story is a bit more complicated at the high school level, as there now is a body of evidence suggesting that students enrolling in small, non-selective high schools are more likely to graduate in four years than their peers in other high schools. We have glimmers of why this might be—the benefits of denser interactions between teachers and students, the fact that students are more likely to take the same courses in small schools than in larger ones, and so forth. But the picture is complicated: students can only enroll in courses that are offered, and small, non-selective high schools may not offer advanced or elective courses available at larger schools. A constrained curriculum is a mixed blessing, offering more equitable access to a school's array of courses, but at the cost of "leveling down" via the absence of the advanced college-preparatory courses providing a platform both for college and career.

A naïve reading of these findings is that we should simply close more large high schools and replace them with small, non-selective schools, sit back, and wait for the graduation rate to skyrocket. But such a policy prescription ignores the fact that New York City high schools are not independent of one another; when a student enrolls in a particular school, that student has not enrolled in any other school. The cumulation of these choices, mediated by the computer algorithm which matches students with schools, may maximize the probability that a student will be matched with one of his or her top three choices, but it may also result in high concentrations of low-achieving students in some schools but not others. It's not clear precisely where the tipping point lies, but it's indisputable that some schools wind up with more than their fair share of low-achieving students.

A possible solution to this problem is to engineer some systemwide "caps" into the high school choice process, so that no school receives an undue concentration of low-achieving students, either via the computer-driven matching mechanism or via subsequent placements of students who arrive "over the transom" and did not participant in the choice process. Though such a provision would likely reduce the percentage of eighth-graders assigned to one of their top choices, it would result in a more equitable distribution of students among the public high schools in New York City.  
 

Aaron Pallas is the Arthur I. Gates Professor of Sociology and Education at Teachers College, Columbia University.Professor Pallas has devoted the bulk of his career to the study of how schools sort students, especially the relationship between school organization and sorting processes and the linkages among schooling, learning and the human life course.