8.6 Million Traffic Stops · 18 States
Black and Hispanic drivers are searched at up to
the rate of White drivers — yet contraband is found less often
Not everyone pulled over faces the same scrutiny. Search rates reveal who police choose to investigate further — and the disparities are stark.
The “outcome test” is powerful evidence: if police search Black and Hispanic drivers more often but find contraband less often, the excess searches can't be justified by higher criminality. It suggests a lower bar for searching minority drivers.
Key Insight
In every state analyzed, both Black and Hispanic drivers are searched at significantly higher rates than White drivers — and in many states, Hispanic drivers face the highest search rates of any group. Yet contraband is found at equal or lower rates for both groups compared to White drivers. This pattern is strong evidence of racial bias in search decisions.
Same car. Same speed. Same road. The only difference? The driver's race. See how outcomes diverge for the exact same stop.
In Arizona, a Black driver is 1.3× more likely to be arrested during a traffic stop than a White driver.
Racial disparities in traffic stops aren't limited to a few bad departments. They're everywhere. Hover over a state to see its data.
Each state tells its own story. Explore the data behind the disparities.
2009-05-28 → 2017-12-31
2009-07-01 → 2016-06-30
2010-01-01 → 2017-12-31
2013-10-01 → 2015-10-01
2010-01-12 → 2018-12-31
2012-01-01 → 2017-12-31
2007-01-01 → 2014-03-31
2007-01-01 → 2015-12-31
2010-01-01 → 2015-01-01
2009-01-01 → 2017-12-31
2002-01-01 → 2016-10-01
2009-01-01 → 2016-12-31
2000-01-01 → 2015-12-31
2010-01-01 → 2017-12-31
2005-01-02 → 2015-12-31
2005-01-01 → 2016-12-27
2010-07-01 → 2015-12-31
2010-01-01 → 2016-05-16
Are things getting better? Track how search rates have changed year by year.
Knowledge is protection. Here's what every driver should know when pulled over.
You must provide your license, registration, and insurance. Beyond that, you have the right to remain silent. You can say: "I choose to remain silent."
If an officer asks to search your car, you can refuse. Say clearly: "I do not consent to a search." They may search anyway if they have probable cause, but your refusal protects you legally.
You have the right to record police interactions in all 50 states. Keep your phone visible and don't interfere with the officer's duties.
Keep your hands visible. Don't reach for anything until asked. Turn on your interior light at night. Your safety comes first — assert your rights calmly.
Note the officer's name, badge number, patrol car number, and agency. Write down what happened as soon as possible. File a complaint if your rights were violated.
If arrested, say: "I want to speak to a lawyer." Do not sign anything or make decisions without legal counsel.
Data alone doesn't create change. Share this evidence, contact your representatives, and support organizations fighting for reform.
New data, updates, and analysis from Justice Index.
This project uses data from the Stanford Open Policing Project, which collected and standardized over 200 million traffic stop records from across the United States.
E. Pierson, C. Simoiu, J. Overgoor, S. Corbett-Davies, D. Jenson, A. Shoemaker, V. Ramachandran, P. Barghouty, C. Phillips, R. Shroff, and S. Goel. “A large-scale analysis of racial disparities in police stops across the United States.” Nature Human Behaviour, Vol. 4, 2020.
We analyze search rates, hit rates (the “outcome test”), and yearly trends across 17 states with the most complete data, from coast to coast. Search rates measure how often stopped drivers are searched. Hit rates measure how often those searches find contraband. The outcome test — if a group is searched more but contraband is found less often — provides evidence of bias.
We run logistic regressions for each state to control for confounding variables. The search model predicts whether a stopped driver is searched, controlling for age, sex, year, stop type, and violation category — with race as the key independent variable (white drivers as reference). We report odds ratios: a value of 2.0× means that group is twice as likely to be searched after accounting for all other factors. The arrest model uses the same controls to predict arrest. The hit rate model tests the “outcome test” — among searched drivers only, whether contraband is found at different rates by race. Lower hit rates for groups searched more frequently suggests the higher search rate is not justified by contraband discovery.
Models use up to 100,000 observations per state (randomly sampled where datasets are larger). Confidence intervals are 95%. Statistical significance is indicated by ★ markers (★★★ = p < 0.001).
Justice Index · Three Investigations
Bias doesn't stop at traffic stops. It follows people from the street to the courtroom to the bank.