What the data shows
Each analysis is its own page: one finding, the method behind it, and every caveat that comes with it. Published dates say when an analysis shipped; the numbers are computed live from the latest MPD release, so the data window on each page is what dates the figures.
Stop disparity is worst where the fewest Black people live
MPD stops, 2023–2025 (pooled) · Published July 2026
6.67× The disparity over timeThe disparity persisted, and deepened, across six years, two mayors, and a pandemic
MPD stops, Jul 2019–2025, by year · Published July 2026
1.73→1.87× Searches: the outcome testSearches find contraband at the same rate by race; who gets searched at all is what differs
MPD stops, both eras (2019–2025) · Published July 2026
up to 4.5× Arrests by districtArrests show the same shape as stops, deeper and starker: thirteen years of it
MPD adult arrests, 2013–2025 (pooled) · Published July 2026
8.08× Camera vs officer enforcementWhere officers choose to write traffic tickets is not where cameras record violations
MPD moving violations, 2023–2025 (36 months, pooled) · Published July 2026
up to 2.6× Veil-of-darkness testThe veil-of-darkness test: does daylight, when race is visible, change who gets stopped?
MPD stops, 2019–2025, inter-twilight window · Published July 2026
OR 0.93 Enforcement vs reported crimeAgainst the crime residents report: stop and arrest volume roughly track violence; frisks exceed every baseline
MPD incidents vs stops (2023–2025) and arrests (2013–2025) · Published July 2026
frisks 2.7× The search-threshold testThe threshold test: does it take less evidence to get searched if you are Black?
MPD stops 2023–2025, searches with recorded outcomes · Published July 2026
Δt +0.038 Classification mix over timeData reliability: did losing the crime-review unit in 2013 leave a fingerprint in the offense mix?
MPD reported incidents, 2008–present, monthly, citywide · Published July 2026
6/12 shiftedAnalyses are listed in the order they shipped. Findings state what the data shows relative to residential population, not a proven cause.