A Missourian Special Report


In 1999, national polls found that more than half the public believed police actively engaged in racial profiling. Three in four blacks believed profiling occurred regularly.

In response, the 2000 Missouri Senate devised a method of determining whether police treated citizens differently, based on the citizen's race. The result was an annual report published by the attorney general which ostensibly measured how fairly races were represented in traffic stops.

Far from resolving the issue, the state's report highlighted consistent disparities for blacks in many areas of the state, including Columbia.

Some say the report is solid proof of biases they've long known were present in policing.The police dismiss the report as unfair. Others aren't sure what to make of it. Many have never heard of it. No one agrees what to do with it.

The Missourian conducted a two-part analysis to help clarify what the numbers really say about race and policing.The first dug deeper into the 2012 traffic stop statistics to see what evidence there really is for bias against blacks in Columbia. The second looked at profiling as a geographic issue to see if police tend to focus on particular neighborhoods, which could disadvantage blacks living in those areas.

The purpose of this study is not to determine the motives of individual officers, whether prejudiced or not, nor does this study indicate the likelihood of blacks to be law-abiding or not.

Instead, this study looks at whether blacks are fairly represented in traffic policing by simple statistical measures and, if not, tests several possible explanations in hopes of encouraging constructive and informed conversation about the issue.

Racial Disproportions

What's fair?

The Missourian's analysis used census data for the driving age population to compare the proportion of black drivers stopped by cops to the proportion they represent in the community. In several categories, the disparity for blacks was significantly worse than has been reported by the attorney general.

Use the following charts to compare these disparities. You may mouse over the outer chart circle to see the number of stops in each traffic stop category. Also, remember a disparity index of 1.0 is considered fair representation.

Chart Key

Under Suspicion

Blacks were four times overrepresented in investigative stops during 2012.

Pre-stop justifications are reasons cited by an officer for pulling a car over. The Missourian grouped these reasons into either moving or non-moving violations.

Investigative stops are a special category assigned by the Columbia Police when a driver is detained in connection with an investigation or observed under suspicious circumstances. Investigative stops don't necessarily indicate the driver has committed a traffic violation.

There were 393 investigative stops made last year.

(It's also important to remember that an officer can indicate multiple reasons for stopping a car.)

Disparity Indexes:

  • Investigative: 4.13
  • Moving: 1.95
  • Non-moving: 2.47
>>Jump to the story
Pre-Stop Justifications
 Black   Other   White 

Under Arrest

Blacks were also four times overrepresented in arrests made as a result of traffic stops.

While blacks represented 10 percent of the driving-age population in 2012, they made up 44 percent of all traffic stop arrests.

There were 1,061 traffic stop arrests made last year.

Drug-related arrests following traffic stops also show large disparities for blacks, who made up 45 percent of the total drug arrests.

Disparity Indexes:

  • Arrests: 4.36
  • Citations: 2.04
  • Warnings: 2.25
>>Jump to the story
Stop Result
Drug Arrest
 Black   Other   White 

State Report: Margin for Error

In its annual report, the attorney general's office has consistently reported blacks are two times overrepresented in traffic stops made by the Columbia Police.

The methodology used to compute that measure has been criticized by members of the law enforcement and research communities for using residential population as a baseline for determining racial bias. A major flaw, they say, is that motorists stopped while passing through a community where they do not live can skew local statistics and inflate the measure of racial bias.

A study in 2003 compared residential populations to an estimated driving population, which took into account passers-by. Using the driving population as a baseline, the researchers found the 2003 disparity index dropped by 6 percent or .13 points for blacks.

Using the 2003 study as the best estimate of the unknown difference between residential and driving populations, the disparities cited in investigative stops and traffic stop arrests far outweigh a margin for error.

>>Jump to the story
State Annual Racial Profiling Reports - All Traffic Stops
Select a report year:
 Black   Other   White 

The Geography of Race and Policing

Traffic stops by the Columbia Police Department in 2012. Stops are concentrated within the First Ward.
Moran's I = .37
Black population as a percentage, based on 2010 census data. The First Ward has highest the number of blacks of all wards.
Moran's I = .73
White population as a percentage, based on 2010 census data.
Moran's I = .54
Stops for black drivers are most concentrated in the First Ward. Black drivers are consistently overrepresented in traffic stop statistics.
Moran's I = .46
Stops for white drivers are also concentrated in the First Ward, though whites make up a smaller proportion of the population in the ward. This supports the argument for geographic or neighborhood-based bias.
Moran's I = .34
Investigative stops are concentrated in the First Ward. Blacks are heavily overrepresented in this category of stops, making up 41.5% of the total. They represent 18% of the First Ward's population.
Moran's I = .29
Calls for service to the Columbia Police are concentrated in the First Ward. These were found to be the most consistent predictor of traffic stops in an area out of the variables the Missourian considered.
Moran's I = .59
Traffic volumes are dispersed and do not predict the concentration of stops in the First Ward.
Moran's I = .68
Population density is dispersed and does not predict the concentration of stops in the First Ward.
Moran's I = .67
Concentration Key:
 High   Low  Outlier
Select a layer from the categories at left to see high and low concentrations and explore possible correlations between data layers.

    When the Columbia Police Department evaluates crime, they make maps that show hotspots around the city. These maps show where crimes are concentrated and can be compared to find correlations, for example, that most break-ins happen in lower-income neighborhoods.

    Taking a page from the crime analyst's book, the Missourian created the map at left to compare several different variables, which might contribute to the number of traffic stops in an area.

    The map's layers were created using geo-statistical software to analyze high and low clustering, that is, areas or neighborhoods where a variable shows markedly high or low values. (Outliers represent a low area neighboring a high cluster, or vice versa.)

    For example, when selecting the traffic stops layer at left, the cluster analysis shows that most of the high numbers of traffic stops are concentrated in the First Ward

    Black population is also concentrated in the center of the city and into the Second and Third wards.

    High concentrations of stops for both black and white drivers were located in the First Ward. This supports the argument that a geographic or neighborhood-based bias is a more significant factor than a race-based bias against the driver.

    Investigative stops are heavily concentrated in the First Ward.

    Calls for service are made to the Columbia Police by callers in the community. According to the police department, where police are asked to respond to these calls determines in large part where traffic stops are made.

    Spatial regression, another geo-statistical analysis technique, was used to test that claim and found that areas with high calls for service were consistently correlated with areas that had high traffic stop numbers. That finding corresponds to the map, but calls for service only told half the story in the regression analysis, only explaining 48 percent of the difference in the number of stops between areas.

    Other factors don't explain very well where the highest number of traffic stops were made.

    Local traffic volumes were evaluated using MODOT's most recent average daily traffic survey. The cluster map shows high concentrations of traffic volumes do not correspond neatly to traffic stops. Traffic policing doesn't appear to focus on those areas where most people are driving.

    Overall population density also does not appear to correlate to traffic stops.

    Want to read more about how we did our analysis? Click the specs.