Minneapolis Police Stop Data: Racial Disparity Analysis

View full report

Problem

Do Minneapolis police stop, search, and arrest drivers at different rates based on race? Using 2017-2020 traffic and suspicious vehicle stop data, this analysis examines whether racial disparities exist in policing patterns and whether those disparities suggest differential treatment by officers.

Approach

Methods: Benchmark test (comparing stop rates to population demographics) and outcome test (analyzing search success rates by race)

Data sources:

• Minneapolis Police Stop Data (2017-2020): ~100,000 stops of vehicles
• U.S. Census/ACS 5-year estimates (2019): Minneapolis demographic data

Key techniques:

• Benchmark test: Calculated stop rates per capita by race, comparing to residential population shares
• Outcome test: Analyzed "hit rates" (citation and booking rates after searches) to see whether search decisions showed different thresholds by race
• Geographic analysis: Examined disparities across Minneapolis neighborhoods

Tools: R (tidyverse, tidycensus, sf, ggplot2)

Key Findings

Disproportionate stops:

• Black drivers stopped at 5x the rate of white drivers relative to population
• Native American drivers stopped at 2.5x the rate of white drivers
• Latino drivers stopped at slightly higher rates (12% vs. 11%)

Higher search rates for minorities:

• Black drivers 3x more likely to have vehicle searched or be frisked than white drivers
• Native American drivers 5-6x more likely to be searched

Lower "hit rates" signal discrimination:

• When searches occurred, white drivers were booked 45% of the time
• Black drivers were booked only 32% of the time
• Latino drivers were booked 36% of the time
• Pattern held across most neighborhoods, particularly those with the highest stop volumes

Interpretation: Lower booking rates for Black and Latino drivers after searches suggests officers applied lower evidentiary thresholds when deciding to search minority drivers—searching on less evidence compared to white drivers.

Files

• mpls_stop_analysis.Rmd - Full analysis notebook with methodology
• mpls_stop_analysis.html - Rendered results with visualizations
• Data sourced via API from Minneapolis Open Data Portal

Context

Analysis applies methodology from Stanford's Open Policing Project to Minneapolis stop data.

Assumptions/caveats: Stop and search rates alone don't prove bias, as they don't account for underlying differences in traffic violation rates or transportation patterns. However, the outcome test (differential hit rates) provides stronger evidence of disparate treatment, as it controls for the decision to search.