This paper examines the impact of a gifted program on retention in an urban school district using a regression discontinuity design. Gifted programs often employ IQ thresholds for admission, with those above the threshold being admitted. One common problem with the RD design arises if the forcing variable (the IQ score) is manipulated, thus invalidating the standard research design. We proposed a modified RD estimator that deals with manipulation in the forcing variable. Once we properly correct for manipulation of test scores around the cut-off points, we find evidence that the gifted program offered by the district has a positive effect on retention of higher income students.