SFPD unveils new disparity benchmarks and live stops dashboard; methodology and data quality remain under review

San Francisco Police Department staff presented a new set of disparity benchmarks and a live stops dashboard to the Police Commission on June 11, aimed at providing more nuanced context for stop data beyond the traditional census-based comparison.

Catherine McGuire, executive director of the Strategic Management Bureau, said the department published a stops dashboard on the department website and plans to add use-of-force and arrests dashboards. Captain Sean Frost described the new benchmarks: the existing census benchmark (using American Community Survey residential data), a risk-adjusted disparity index (RAD, which compares stop contacts to violent-crime victimization risk by race), a suspect-adjusted disparity index (SAD, which compares stops to reported suspect demographics) and a traffic analytic method that uses not-at-fault collision drivers as a proxy for the driving population.

Frost said the department ran the RAD and SAD analyses for six quarters of stop data. He summarized preliminary findings: the RAD comparison of Black to White populations was near parity across the period analyzed; RAD suggested Hispanics were stopped less frequently than their risk of violent-crime victimization would predict; SAD suggested Black and Hispanic populations were underrepresented in stops relative to the race of reported violent-crime suspects. The department posted code and details online and said the QUADAR (quarterly activity and data report) includes a fuller explanation.

Commissioners and public commenters raised concerns about data quality. Commissioner Benedicto reminded the commission of a DPA audit that identified discrepancies between dispatch records and stop data entries; department staff said the particular errors identified in that audit did not affect racial-demographic counts used in the new benchmarks and that the department’s error rate for the relevant fields had dropped below 5%. Director Henderson and DPA staff said additional audits and improvements remain ongoing and that more detailed validations are part of an evolving process.

Department staff emphasized that no single benchmark is definitive and that the new measures are intended to supplement, not replace, the census benchmark. They said the expanded set of metrics is meant to give the public and the department more context when interpreting stop statistics and to help inform future policy and accountability work.