Representative bureaucracy theory examines how bureaucrats’ demographics impact outcomes for clients with shared identities, with “critical mass” posited as an enabling condition. Yet empirical evidence is mixed regarding where this threshold stands. To reconcile these inconsistencies, this study emphasizes the need to first clarify the mechanisms that underpin critical mass requirements. Specifically, I attend to how majority behavior changes due to enhanced representation and evaluate corresponding critical mass condition. Nonparametric analyses of traffic stop data in two states find that, in Washington, the critical mass where White officers show reduced bias towards Black drivers occurs when Black officers constitute 6-9 percent of the force. In South Carolina, similar shifts occur at 9-11 and 19-23 percent Black representation. While findings indicate improved policing towards Black drivers, increased representation still falls short of achieving full parity between Black and White drivers. No significant critical mass is observed for Hispanic representation in either state.