Does Sorting Students Improve Scores? An Analysis of Class Composition

By Amber M. Winkler

THE EDUCATION GADFLY WEEKLY

A wise wonk once wrote that the biggest challenge facing America’s schools is the enormous variation in the academic level of students coming into any given classroom. The subject of this NBER working paper is one proposed solution to this quandary: sorting students by ability. And though conventional wisdom (and some prior research) suggests that kids in the lower-achieving groups would fare worse with such an approach, the researchers in this study concluded that sorting is beneficial for both high and low achievers—though high achievers did see larger gains than those of their lower-scoring peers (approximately 1.6 times greater). The analysis used student data linked to one cohort of Dallas elementary students—amounting to roughly 9,000 children in 135 schools who start out in the third grade and end up in the fourth (in 2003–04 and 2004–05). After attempting to account for unobservable ways that schools might sort (say, by student behavior), researchers found that three-quarters of the schools organize students along at least one dimension: Nineteen percent by prior math scores, 24 percent by prior reading scores, 28 percent by “gifted” status, 57 percent by LEP (limited English proficiency) status, and 13 percent by special-education status (further, around 40 percent sort by at least two dimensions). If schools began perfectly grouping by ability, they would see a 0.4 SD gain in student learning—a figure larger than the black-white achievement gap. But while the argument for increased test scores may be clear, school leaders must still bear in mind goals other than those on which this study focused, including things like classroom culture and social/emotional skills acquisition.