This analysis is based on police traffic stops data from 12 cities across the country. Since data from each city has its own weaknesses, the analysis for each city is slightly different as well. But here are concepts that appear in most cities and thus worth taking a look at before jumping to your city.
Two city datasets are confined to stops initiated for criminal enforcement while the others also include stops made for other purposes like traffic enforcement. Boston and New York have criminal enforcement data only, which is called “Stop-and-Frisk” data in New York, and “Field Interrogation and Observation (FIO)” data in Boston.
Although most city datasets include stops made for various reasons, few specify the reason for each stop. Therefore, this analysis doesn’t examine differences between stops made for different purposes, but I marked cities with that information for those who want to take a deeper dive.
Some city datasets have information on the subject type - whether the subject was a pedestrians or driver. If the analysis for your city is not diveded into the two parts - pedestrian and vehicular stops - that means your city data doesn’t have that information. You can find more details in your city section.
Police officers may follow up a stop with a frisk (or pat-down) if they have reasons to suspect that the person is armed and dangerous. A frisk is supposed to be a limited search of the person’s outer clothing for weapons.
A further search, on the other hand, involves officers probing for any kind of evidence - whether it be weapons, drugs or stolen items. Officers can search a vehicle without a warrant or consent if they have probable cause to believe there is evidence of a crime in your vehicle or reasonably believe a search is necessary for their own protection - a hidden weapon, for example.
It would be interesting to see if minority drivers were searched without a warrant or consent at a higher rate than whites, but unfortunately, most data doesn’t have that information.
In this analysis, the frisk and search rates refer to the proportion of stops followed by a frisk and search, respectively. These are among the key concepts that measure and reveal police officers’ implicit racial biases because frisks and searches often involve officers’ second judgement.
If a certain race group is frisked or searched at a much higher rate, that means officers feel that group is more suspicious or dangerous than others.
Another key concept is the hit rate, which is the proportion of successful frisks or searches where any kind of contraband was found. Although it doesn’t appear in all city analyses, the hit rate is another important concept that reveals officers’ implicit biases.
If the hit rate for a certain race group is much lower than that for the other groups, it means that group has more innocent people being frisked or searched.
Similar to the frisk and search rates, the summons and arrest rates are the proportion of stops led to a summons issuance and arrest, respectively.
Some city agencies stop pedestrians and drivers more aggressively than others, but regardless of their aggressiveness, racial disparities are found in most cities. Blacks and Hispanics were more likely to be stopped than whites in all the cities, except for Durham, NC. And once stopped, blacks face a frisk or search at a higher rate than whites. Within the same race, males are more likely to be stopped and searched than females.
Subjects whose race is known but ethnicity is unknown are categorized as their non-Hispanic race group because officers stopped them without any knowledge on their ethnicity. This allowed me to contain as many stops as possible.
City: Philadelphia, Pennsylvania (GEOID: 4260000)
Population: 41% Black, 35% White, 14% Hispanic
Data Source: Philadelphia Police Department (PPD)
Data Timeframe: Jan. 1, 2014 - May 23, 2020
Stop Type: Pedestrian and Vehicular (2,745,157 stops in total - 864,445 pedestrian and 1,880,711 vehicular stops)
Data Strengths or Weaknesses: Less than 1 percent of stops don’t have racial information on the suspects. Data doesn’t have information on the reasons for stops and further searches.
The PPD stopped nearly 3 million pedestrians and drivers over the past six years. In a city where only 41 percent of the population is black, 70 percent of those stopped were blacks, and most of them were males. That means blacks are 3 times more likely to be stopped than whites.
Over the past six years, the PPD has drastically reduced pedestrian stops and expanded vehicular stops. But racial disparities in both types of stops grew.
In 2014, 61 percent of drivers the PPD stopped were blacks, and now 76 percent are blacks. The share of black drivers among all subjects has increased while that for other races decreased.
Also, Over the past six years, the PPD stopped more black female drivers than white female drivers. This is a unique trend that can’t be found in other cities. The gap has grown bigger and bigger and now the number of black female drivers the PPD stopped double the number of their white male counterparts.
Pedestrian stop numbers show even bigger racial disparities.
As mentioned above, the number of PPD pedestrian stops has drastically declined during the timeframe, but the racial disparities between blacks and whites grew. Black male and female combined, accounted for 69 percent of all pedestrian stops in 2015. The proportion of blacks out of all PPD pedestrian stops has constantly increased - especially black males - and now they account for 73 percent, in a city where only 42 percent of the population is black. And the vast majority of black pedestrians the PPD stopped are black males.
One in three black male drivers and pedestrians the PPD stopped were 25 or younger.
Once stopped, hispanic and black males are nearly twice more likely to be frisked than whites. Blacks and Hispanic males stopped by the PPD face a frisk at 10 percent chances while white males get frisked at only 5 percent chances. Minority males are also more likely to be searched.
Black males have the highest chances of being frisked or searched once stopped but have the lowest hit rate. White females, have the highest hit rate. One in five white females the PPD frisked or searched possessed contraband. That means innocent blacks are being frisked or searched at the highest rate.
Despite having a lower hit rate, Hispanic and black males are more likely to be arrested than white males.
City: Los Angeles, California (GEOID: 0644000)
Population: 9% Black, 28% White, 49% Hispanic
Data Source: Los Angeles Police Department (LAPD)
Data Type: Both Pedestrian and Vehicular (2,047,560 pedestrian and 5,769,833 vehicular stops)
Data Timeframe: 2010 - 2020
Data Strengths and Weaknesses: All stops have racial information of the subject. But no information on what kinds of post-stop activities were conducted and what were the results. So no hit, citation or arrest rates.
Over the past 10 years, the LAPD stopped nearly 8 million pedestrians and drivers. And 25 percent of those stopped are black in a city where only 9 percent of the population is black. Whites make up 28 percent of the population but only account for 21 percent of those stopped by the LAPD. That means blacks are four times more likely to be stopped than whites, and three times more likely than Hispanics.
The LAPD vehicular stops have declined over the past 10 years but racial disparities grew. The proportion of black drivers among all drivers the LAPD stopped has significantly increased.
In 2010, black males accounted for only 11 percent out of all drivers stopped by the city police that year, and that was much lower than the proportion of white males, as it should considering the city population. But the number has dramatically increased and now the number of black male drivers the LAPD stops is twice the number of white male drivers stopped, in a city where white population is three times more bigger that black population.
Black males make up even a bigger share when it comes to pedestrian stops. This is a trend that can be found in other cities, including Philadelphia.
More than one in four pedestrians the LAPD stopped over the past six years are black males.
As mentioned earlier, this data has information only on whether a stop followed by a further activity, but not exactly what kind of activity it was.
Once stopped, blacks and Hispanics are much more likely to face a further activity - whether it be a frisk, search or arrest. Nearly 40 percent of black males stopped face a further activity while less than 20 percent of white males do so.
A trend unique in Los Angeles is that Hispanic and Black females are more likely to face a further activity than white males. In most cities, males are more likely to be frisked or searched, regardless of their race. Also, black females are twice more likely to face a post-stop activity than white females. Need to reach out to the LAPD to get information on what exactly “post stop activity” means.
City: Minneapolis, MN (GEOID: 2743000)
Population: 19% Black, 60% White, 10% Hispanic
Data Source: City of Minneapolis
Data Timeframe: November 2016 - Current Date
Data Strengths or Weaknesses: No distinction between pedestrian and vehicular stops. Also, doesn’t have information on search result - whether contraband was found or the individual was arrested. Just whether individuals or their vehicles were searched or not. Moreover, 24 percent of stops don’t racial information of the subject.
A good thing is that this data has reasons for stops. The six distinctive reasons for stops follow, in the order of their share, from the most to the least:
The Minneapolis Police Department (MPD) has made 161,644 stops since November, 2016. More than half of the subjects whose racial identity is known, are black and 32 percent are whites in a city where only 19 percent of the population is black. Whites account for 60 percent of the population but take up only 32 percent of the traffic stops. That means blacks are five time more likely to be stopped than Hispanics and whites.
The number of MPD stops has drastically declined for the past three years, but racial disparities grew. The share of black males among those stopped, has constantly increased while that of white males dropped. Now black males alone account for nearly half of all individuals the MPD stops.
Black males have higher chances of getting searched after being stopped. One in four black males the MPD stops face a further search while only one in 10 white males the MPD stops face a search. Also, black females are more likely to be searched than white males.
City: San Francisco, California (GEOID: 0667000)
Population: Black 5%, White 41%, Hispanic 15%
Data Source: San Francisco Police Department (SFPD)
Data Type: Vehicular
Data Timeframe: 2014-2017
Data Strengths or Weaknesses:
The SFPD made less than 1 percent - 0.1 percent specifically - of the stops with a warrant, and 99 percent of stops were based on traffic violations. Below are the unique reasons for stops in the order of their share, from the most to the least:
The SFPD made nearly half a million traffic stops - 442,004 stops, specifically - between Jan. 1, 2014 to Dec. 31, 2017. In a city where only 5 percent of the population is black, 15 percent of those the SFPD stopped are blacks while whites that make up 41 percent of the population, account for 37 percent. That means blacks are three times more likely to be stopped than whites in the city.
The share of minority drivers, especially black males, has increased during the timeframe. In 2014, black males accounted for 10 percent of individuals stopped by the SFPD, but the number has constantly increased and now they make up 13 percent. During the same timeframe, the proportion of Hispanic male drivers has gone up as well from 10 percent to 12 percent.
One unique trend in San Francisco that can’t be found in other cities, is that minority females have more chances to be searched once stopped, than white males. This is disturbing when considering that most SFPD stops are initiated by traffic violations. As you can see, black females have higher chances to be searched than Hispanic and white males. Hispanic females are more likely to be searched than white males.
When compared within the same sex, black males are seven times more likely to be searched than white males, and black females are six times more likely to be searched than white females.
Although whites face a post-stop search at much lower percent chances than blacks, they get arrest at higher chances. 75 percent of white males stopped by the police were arrested while only 55 percent of black males stopped by the police were arrested. That might mean the SFPD is stopping males only when necessary. Also note that in other cities, black male subjects are more likely to be arrested than white males.
Despite having the lowest arrest rate, black males have the highest citation and warning rates. Click on the name of the column by which you want to sort the table.
City: Louisville, Kentucky (GEOID: 4743820)
Population: Black 5%, White 90%, Hispanic 1%
Data Source: Louisville Metro Police Department (LMPD)
Stop Type: Vehicular (for traffic enforcement)
Data Timeframe: 2009-2019
Data Strengths or Weaknesses: One very unique strength of Louisville data is that it include demographic information - race and age - of police officers as well. But this analysis doesn’t examine the dynamics between the race of POs and drivers. This data also contain the reasons for searches but there was no clean, consistent classification so I didn’t use that information.
Almost all LMPD stops were initiated by the driver’s traffic violation. Less than 1 percent of LMPD stops were made with a warrant.
The LMPD made half a million stops - 415,907 stops specifically - since 2009. 27 percent of those the LMPD stopped are black. And blacks make up only 5 percent of the population is black. Whites make up 90 percent of the city population but account for only 68 percent of the stops. That means blacks are seven times more likely to be stopped than whites.
The number of LMPD stops drastically declined over the past six years, but racial disparities grew. The proportion of black male and female drivers has increased while that of white male and female drivers dropped.
Blacks and Hispanics drivers have a higher warning rate and lower citation rate than white drivers. That means, minority drivers are more likely to be stopped when they didn’t commit traffic violation serious enough for an officer to issue a citation.
City: Houston, Texas (GEOID: 4835000)
Population: 23% Black, 57% White
Data Source: Stanford Open Policing Project
Data Type: Vehicular Stops Followed by Citation Only
Data Timeframe: Jan. 1, 2014 - Apr. 8, 2018
Data Strengths or Weaknesses: Ethnicity is not considered, so blacks and whites here include Hispanic blacks and whites, respectively. Also, 20 percent of the subjects don’t have racial information. All stops in the data are those followed by a citation. No arrest and contraband recovery data. So no citation, arrest or hit rates here. Data has information on reasons for stops but the column was dirty so I didn’t dig deeper.
The HPD
In a city where only 23 percent of the population is black, blacks account for 37 percent of those stopped over the past few years. That means blacks are 1.5 times more likely to be stopped in the city.
City: Nashville-Davidson Metropolitan Government, TN (GEOID: 4752006)
Population: 28% Black, 55% White, 10% Hispanic
Data Source: Stanford Open Policing Project
Data Type: Vehicular Stops for Traffic Enforcement
Data Timeframe: December 2009 - March 2019
Data Strengths or Weaknesses: Less than 1 percent of stops don’t have the racial information of those stopped. This data has reasons for stops.
Nashville is one of the areas with smaller racial disparities. In a city where 28 percent of the population is black, nearly 40 percent of those stopped were blacks. Blacks were only 1.4 times more likely to be stopped than whites.
Although blacks and whites are stopped at a similar rate in the area, blacks are much more likely to be frisked or searched, once stopped. Once stopped, black males are 3 times more likely to be frisked, and 2 times more likely to be searched than white males. Black females are more likely to be frisked and searched than white females as well.
Blacks were more likely to be arrested than whites as well.
City: New York, New York (GEOID: 3651000)
Population: 22% Black, 32% White, 29% Hispanic
Data Source: New York Polie Department (NYPD)
Data Type: Stop-and-Frisk for Criminal Enforcement
Data Timeframe: 2015 - 2019 (data available from 2003)
Data Strengths and Weaknesses: This data is a gold mine with information on what kinds of force officers used when conducting a stop if they used any. Frank needs to take a look at this. One percent of suspects’ racial information is unknown.
In a city where only 22 percent of the population is black, 56 percent of those stopped by the NYPD’s stop-and-frisk program Blacks are 8 times more likely to be stopped than whites in New York.
The NYPD significantly significantly reduced the number of suspects they stop in 2016 and maintained a similar level ever since, but the racial disparities between whites and blacks have grown. The share of black males among all stop suspects have increased since 2015 while that of white males decreased.
Also note that the percentages below are based on 94 percent of stops where racial information of the suspect is known.
Black females and males had the lowest hit rates. White females had the highest hit rates. Almost one in five white females frisked or searched possessed contraband while one in eight blacks searched possesed contraband.
City: Raleigh, North Carolina (GEOID: 3755000)
Population: 28% Black, 53% White, 11% Hispanic
Data Source: North Carolina State Bureau of Investigation (NCSBI)
Data Timrframe: 2000-2019
Data Strengths or Weaknesses: This data is a gold mine as well. It has so much information other data doesn’t have, including specifics on contraband - how many ounces of drug was found, for example. It would be interesting how many black juveniles have been arrested for possessing a small amount of week. I’ve finished cleaning the data so I’ll send it to Samantha later. Only 1.2 percent of stops don’t have racial information on those stopped.
Only 0.2 percent of searches conducted in Raleigh and Durham PDs were conducted with a warrant.
## # A tibble: 5 x 3
## search_type count pct
## <chr> <int> <dbl>
## 1 Consent 17896 39.5
## 2 Search Incident to Arrest 13583 30
## 3 Probable Cause 12283 27.1
## 4 Protective Frisk 1473 3.2
## 5 Warrant 90 0.2In Raleigh, black males have the highest chances of getting searched, once stopped. Black males are two times more likely to be searched than white males.
In Raleigh, blacks have the highest chances of possessing contraband, once frisked or searched.
City: Durham, North Carolina (GEOID: 3719000)
Population: 39% Black, 39% White, 14% Hispanic
Data Source: North Carolina State Bureau of Investigation (NCSBI)
Data Timrframe: 2000-2019
Data Strengths or Weaknesses: This data is a gold mine as well. It has so much information other data doesn’t have, including specifics on contraband - how many ounces of drug was found, for example. It would be interesting how many black juveniles have been arrested for possessing a small amount of week. I’ve finished cleaning the data so I’ll send it to Samantha later. Only 1.2 percent of stops don’t have racial information on those stopped.
In Durham, whites are nearly twice more likely to be stopped than blacks. Durham is the only city where whites are twice more likely to be stopped than blacks.
White male drivers
City: Boston, Massachusetts (GEOID: 2507000)
Population: 45% White, 23% Black, 20% Hispanic
Data Source: Boston Police Department (BPD)
Data Type: Both Pedestrian and Vehicular for Criminal Enforcement
Data Timeframe: 2015 - December 2019 (Data available from 2011)
Data Strengths and Weaknesses: The BPD changed its record management system twice - first in 2015 and once more in September 2019 so data stored through three different systems is slightly different.
One big issue in data from 2015 to September 2019 is that only 53 percent of stops’ search information is accurate due to the city’s flawed record management system back then. So the frisk and search rates by race and gender from the timeframe is based on the 53 percent of the stops. Also, about 4 percent of stops don’t have racial information on those stopped. Also, no information on whether contraband was found or not. Just whether a summons was issued.
But this data has information on 1) circumstances of field contact (Observed, Encountered, Stopped), and 2) basis for field contact (Encounter, Intel, Reasonable Suspicion, Probable Cause).
In a city where only 23 percent were black, about 65 percent of those investigated were blacks, and 92 percent of them were males. That means blacks are 7 times more likely to be stopped than whites in the city.
Again, frisk search rates for the BPD are calculated based on 54 percent of stops with accurate search, frisk and summons information.
There are little racial disparities in frisk and search rates. Once stopped, males are frisked at a 25 percent chance, regardless of their race and ethnicity.
But the summons rate - the percentage of those later summoned - shows racial disparities. Whites have a larger share of those who later summoned than blacks, which means officers are stopping more innocent blacks than whites.
City: Chicago, Illinois
Data Source: Illinois Department of Transportation (IDOT)
Data Type: Both Pedestrian and Vehicular
Data Timeframe: Jan. 1, 2014 - Dec. 31, 2018 (Pedestrian stops data is available from 2017, which is the first year the data was collected.)
Data Strengths or Weaknesses:
There are four distinctive reasons for stops: Moving Violations, Equipment, License Plate/Registration and Commercial Vehicle. And there are six types of moving violations: Speed, Lane Violation, Seat Belt, Traffic Sign or Signal, Follow Too Close and Other.
In this analysis, I included all kinds of stop. The number of vehicular stops constantly increased over the past five years and so does the racial disparity between blacks and whites.
Blacks are four times more likely to be stopped than whites.
Between 2014 and 2018, racial disparities in vehicular stops grew significantly. the share of black male and female drivers surged while that of white male and female drivers plummented.
No arrest rate. For whatever reason, can’t see search rate. Couldn’t figure out after spending 30 min. Will fix it later.
## # A tibble: 6 x 5
## # Groups: clean_race [3]
## clean_race gender warning_rate citation_rate search_rate
## <chr> <chr> <dbl> <dbl> <dbl>
## 1 Black Female 77.0 23.0 0
## 2 Black Male 76.3 23.7 0
## 3 Hispanic Female 69.8 30.2 0
## 4 Hispanic Male 64.6 35.4 0
## 5 White Female 63.2 36.8 0
## 6 White Male 62.9 37.1 0City: New Orleans, Louisiana
Population: 59% Black, 31% White, 6% Hispanic
Data Source: New Orleans Police Department (NOPD)
Data Type: Both Traffic and Criminal Enforcement
Data Timeframe: 2016 - 2019
This data has reasons for stops. Below are the reasons in the order of their share, from the largest to the smallest:
In a city where 59% of the popuation is black and 31 percent is white, blacks are 1.4 times more likely to be stopped by the police than whites.