Washington, DC · 2019 to 2025
Daylight doesn't change who DC police stop
If officers stopped people based on race they could see, daylight stops would skew toward Black people. At the same clock time, they don't. The test checks one mechanism, so it doesn't offset the other findings.
If officers decided whom to stop based on race they could see, stops made in daylight would skew toward Black people compared with stops after dark. The test for this (Grogger and Ridgeway, 2006) uses the seasons: 7 p.m. is bright in June and dark in December. Comparing stops at the same clock time, in daylight and after dark, separates what officers can see from who is out at that hour.
The analysis uses 127,947 DC police stops made between 4:45 and 9:09 p.m. from July 2019 to 2025. Black people were 77.8% of daylight stops and 80.2% of stops after dark. At the same clock time, the odds that a stopped person was Black were slightly lower in daylight (odds ratio 0.93, 95% interval 0.90 to 0.96). That is the opposite of what visible-race profiling predicts.
78,104 daylight and 49,843 after-dark stops. 13,940 stops between sunset and dusk, when lighting is ambiguous, are left out. Lighting from NOAA's solar calculations for DC. Source: MPD stops of persons via DC Open Data.
The small difference comes from place, not light
Compared within each police district, daylight and after-dark stops can't be told apart (odds ratio 1.02, not significant). The small citywide difference comes from where stops happen at different times of year, not from lighting. Within 30 days of a daylight-saving clock change, where the season is held nearly constant, there is also no difference (0.98).
One small signal remains. Within districts, ticket-only stops, the closest thing in this data to traffic stops, were slightly more likely to be of Black people in daylight (odds ratio 1.06, 95% interval 1.02 to 1.11). Near the clock changes it can't be confirmed (1.05, 0.94 to 1.16).
What this does not show
- Other kinds of bias. The test checks one mechanism: whether seeing race changes who is stopped. In a brightly lit city, race is often visible after dark too, which weakens the test. Its result neither confirms nor offsets the stop, search and arrest differences in the stops report and the searches report.
- Traffic stops alone. MPD's stop data doesn't separate traffic from pedestrian stops; ticket-only stops are the nearest stand-in.
- Exact group membership. Race is recorded by officers.
- People. Records count stops, not people: someone stopped twice counts twice.
Part of the series Policing in DC, 5 reports.