Can Tracking Raise the Test Scores of High-Ability Minority Students?

District-wide regression discontinuity at the achievement-rank admission margin, multiple cohorts

Grade 4-6 students in a large diverse US district (Broward County FL); separate classrooms for the highest-achieving students in each school, including non-gifted-labelled students.

The same district and the same programme as the gifted-margin nulls, but with the admission margin defined by prior ACHIEVEMENT RANK rather than an IQ label — which makes this the cleanest available label-versus-content decomposition. LATE at the rank margin.

The most important gifted-programmes result for a school builder: separate classrooms selected on demonstrated ACHIEVEMENT produce large, persistent gains for high-achieving minority students at ~zero cost and with no harm to anyone else — while the IQ-based gifted label produces nothing. Select on demonstrated achievement, not labels.

Discontinuity design at the rank margin; the comparison is between students with essentially identical prior achievement, so ability differences are held constant.

Verified including the mechanism-attribution correction. Read together with the same authors' gifted-margin nulls, which is the contrast that carries the conclusion.