2019 VEHICLE STOPS EXECUTIVE SUMMARY
As the chief lawyer for the State of Missouri, my job is to protect each and every one of our six million citizens from crime, abuse and fraud, a responsibility I take very seriously. Our government, the shared responsibility between the citizens of our
state and the elected officials, must be a framework that preserves all citizens’ rights to life, liberty and pursuit of happiness.
The office of the Missouri Attorney General is required, by law, to collect data on the demographics of the traffic stops made by law enforcement officers from across the state, and to report these findings to the Governor and the public. Importantly,
this data can help government and law enforcement determine any issues with disparities related to stops and searches.
This report aggregates the traffic stops data from 571 law enforcement agencies across the state, breaking down the data as it relates to race, the number of stops, the search rate, contraband hit rate and arrest rates. In 2019, we identified several
changes to questions that officers must answer when making a stop that we believe will make next year’s report more informative. This includes questions relating to the officer’s assignment, the residential zip code of the driver stopped
and the reason for issuing a citation or warning. This data will provide more context for the data collected and be fully available in the 2021 report.
As we seek to balance the rights of all citizens of our state with the enforcement of the rule of law, and the brave men and women of law enforcement who put their lives on the line every day to protect us, we will continue to improve this report.
Concerns by the citizens of Missouri and the Missouri legislature regarding allegations of bias in traffic enforcement prompted the passage of Section 590.650, RSMo. SB
1053 created Section 590.650, RSMo. which became effective August 28, 2000. This statute created the Vehicle Stops Report and required that the Attorney General’s Office collect and report on traffic stops conducted by law enforcement officers
across the state of Missouri.
Under § 590.650, RSMo. all peace officers in the state must report specific information, including a driver’s race, for each vehicle stop made in the state. Law enforcement agencies must provide their vehicle stops data to the Attorney General
by March 1, and the Attorney General must compile the data and report to the Governor, General Assembly, and each law enforcement agency no later than June 1 of each year. The law allows the Governor to withhold state funds for any agency that does
not submit its vehicle stops data to the Attorney General by the statutory deadline.
After
reviewing analysis of the Vehicle Stops Report (VSR) and conferring with law
enforcement leaders across the state in 2019, the Attorney General’s Office
(AGO) began implementing comprehensive changes to the VSR. These changes will
improve the information collected for the report while allowing for a
fundamental shift in the level of analysis possible through the VSR. Three new
questions have been added to the report that collect information on officer
assignment during the stop, the residential zip code of the stopped driver, and
the cause of citations and/or warnings issued to the driver. In addition, other
questions have been adjusted for clarity or to improve the value of the data
they collect by adding new response options.
The most significant change to the VSR is its shift toward collecting disaggregated data from across the state. Currently, agencies only report the aggregate numbers of stops meeting the criteria for each question broken down only by race, for example
200 searches of Hispanic drivers over the course of the year. This reporting framework prevents multi-variate analysis with variables like driver age, driver sex, and time of stop that could significantly improve VSR analysis. To correct
this, the AGO is moving to implement an optional data collection framework that collects all variables for each stop an agency made during the year. These changes became effective January 2020 and implementation efforts across the state are ongoing.
The benefits of these changes will begin to manifest in the 2020 report released next year and will be fully realized by the 2021 report. The
benefits of these changes will be fully realized by the 2021 report.
The summary of statewide vehicle stops data has been provided by Dr. Scott H. Decker, Foundation Professor Emeritus in the School of Criminology and Criminal Justice at Arizona State University; Dr. Richard Rosenfeld, Curators’ Distinguished Professor
Emeritus in the Department of Criminology and Criminal Justice at the University of Missouri-St. Louis; and Dr. Jeff Rojek, associate professor in the School of Criminal Justice and Director of the Center for Anti-Counterfeiting and Product Protection
at Michigan State University.
This report summarizes traffic stop data from 571 law enforcement agencies in Missouri for calendar year 2019. An additional 29 agencies indicated they made no traffic stops during the year. These agencies often contract out traffic enforcement to another
agency covering their jurisdiction and focus on other enforcement activities. In
total, this report represents 96.3% of the 623 active law enforcement agencies
in the state.
The agencies filing reports recorded a total of 1,524,640 vehicle stops, resulting in 102,755 searches and 74,553 arrests. Table 1 breaks out the stops, searches and arrests by race and ethnic group. Four summary metrics are included in Table 1 that may
be useful in assessing bias in traffic enforcement.
When assessing these four metrics and the auxiliary statistics alongside them, the concept of discretion is an important consideration. For bias to influence the decisions of law enforcement officers, they must have some discretion in the action they
take. If officers have many possible actions to take in a given situation, then it is possible for bias to affect their decision-making process. However, if officers only have one course of action in a given situation, there is little room for bias
to influence their decisions without breaking from procedure or statute, so caution should be used in attributing bias.
An example of this is when a driver that has been stopped has an outstanding warrant for his or her arrest. In this situation, an officer must arrest and then search that driver. If outstanding warrants become concentrated in a racial/ethnic group in
an agency’s jurisdiction, this may result in the appearance of a disparity in arrest rates when officers had no discretion in their actions and therefore could not have been influenced by bias. Analysis of the degree of discretion that officers
have in various situations is vital for a complete understanding of this report.
| Total | White | Black | Hispanic | Asian | Am. Indian | Other |
---|
16+ Population | 4,730,501 | 3,914,998 | 515,828 | 139,109 | 80,677 | 19,168 | 60,721 |
---|
Stops | 1,524,640 | 1,161,680 | 297,608 | 35,421 | 14,082 | 2,051 | 13,798 |
---|
Searches | 102,755 | 72,387 | 26,371 | 2,940 | 399 | 119 | 539 |
---|
Arrests | 74,553 | 52,873 | 18,472 | 2,244 | 329 | 96 | 539 |
---|
2010 Statewide Population % | 100.00% | 82.76 | 10.90 | 2.94 | 1.71 | 0.41 | 1.28 |
---|
2010 Disparity Index | --- | .92 | 1.79 | .79 | .54 | .33 | .71 |
---|
2018 Estimated statewide population % | --- | 80.42 | 10.98 | 3.38 | 2.03 | .46 | 2.73 |
---|
2018 Disparity Index | --- | .95 | 1.78 | .69 | .46 | .29 | .33 |
---|
Search rate | 6.74 | 6.23 | 8.86 | 8.30 | 2.83 | 5.80 | 3.91 |
---|
Contraband hit rate | 35.37 | 35.93 | 34.93 | 28.37 | 27.82 | 24.37 | 28.76 |
---|
Arrest rate | 4.89 | 4.55 | 6.21 | 6.34 | 2.34 | 4.68 | 3.91 |
---|
Notes: 2010 Disparity index is based on population figures from the 2010 Census for persons 16 years of age and older who designated a single race. Hispanics may be of any race. Other includes persons of mixed or unknown
race. The 2018 Disparity index is based on 2014-2018 average population estimates from the U.S. Census Bureau’s American Community Survey (ACS). The ACS only provides race-specific Hispanic estimates for White, meaning non-White
Hispanic residents are double-counted in the 2018 race percentages above. While the 2010 disparity index is the default metric, if a jurisdiction has a small non-White Hispanic population, the 2018 disparity index may provide more
current information. Disparity index = (proportion of stops / proportion of population). A value of 1 represents no disparity; values greater than 1 indicate over-representation, values less than 1 indicate under-representation. Search rate = (searches / stops) X 100. Contraband hit rate = (searches with contraband found / total searches) X 100. Arrest rate = (arrests / stops) X 100. |
[ back to top ]
1. Disparity Index
The first summary metric is the “disparity index” and it relates each racial/ethnic group’s proportion of total traffic stops to its proportion of the driving-age (16+) population. A value of 1 indicates that a group’s proportion
of vehicle stops equals its population proportion: it is neither “under-represented” nor “over-represented.” Values above 1 indicate over-representation, and those below 1 indicate under-representation in traffic stops.
There are several reasons that traffic stops may not be proportionate to the racial/ethnic composition of a jurisdiction. These reasons often overlap and identifying the magnitude of each reason’s effect or separating them from each other is beyond
the capability of this report or its underlying data. The four major sources of disparity identified by scholars are: [1]
- Policing strategies and policies: Law enforcement officials make strategic choices on where and when to police that may disproportionately impact various racial/ethnic groups. These strategies
may be as simple as concentrating patrols in areas within a city with higher crime rates, but could lead to a disproportionate impact if that area has a higher concentration of a racial/ethnic group than the jurisdiction as a whole.
- Differences in real rates of offending between racial/ethnic groups: The correlation of dynamics such as economic or social disadvantage with race or ethnicity may lead to differences
in rates of real offending. If there are real differences in offending rates, traffic stops should theoretically increase or decrease accordingly.
- Implicit bias: Subconscious or unconscious biases that can influence the decisions and perceptions of individuals. Implicit bias can be difficult to detect, even for the individual operating
under its influence.
- Explicit bias: Overt, conscious bias towards a specific group.
In 2019, the 1,161,680 Whites who were stopped
accounted for 76.2% of all vehicle stops. Whites comprise an estimated 82.8% of
Missouri’s driving age population. Therefore, the value for Whites on the
disparity index is .92 (.762/.828). Blacks represent 10.9% of Missouri’s driving-age
population but 19.5% of all vehicle stops, for a disparity index of 1.79.
Hispanics (0.79), Asians (0.54), American
Indians (0.33), and persons of mixed or unknown race (0.71) were stopped at
rates well below an expectation based upon their portion of the driving-age
population.
The values on the disparity index for the
different groups can be compared directly to one another by dividing their
values. For example, the likelihood that a Hispanic motorist was stopped is 1.46
times that of an Asian motorist (.79/.54).
[1] Lum, C., and Wu, X. (2017). Basic Analysis of Traffic Citation Data for the Alexandria Police Department (2011-2015). Fairfax, VA: Center for Evidence-Based Crime Policy, George Mason University.
Limitations
The disparity index relies upon a benchmark to
set the expected rate at which members of a racial/ethnic group should be
stopped. The actual rate of stops is then compared against this benchmark to
calculate the disparity index. The 2019 Vehicle Stops Report uses the
racial/ethnic portions of the driving-age (16+) 2010 Census populations of each
jurisdiction for its benchmarks. This is an imperfect benchmark as a person
does not need to live in an area to drive through it and a person may not drive
despite being old enough to legally do so. This dynamic can lead to the
racial/ethnic makeup of drivers on a jurisdiction’s roadways differing from
that of its residential population. The extremely low disparity index value for
American Indians, for example, could indicate that they are under-represented
among the state’s motorists. In addition, the benchmark population figures are
nine years older than the vehicle stops data they are compared to. The age of
these estimates may lead to some jurisdictions’ population estimates differing
significantly from their current values.
To minimize the distortion of the disparity
index caused by the use of nine year-old population benchmarks, this year’s
report uses population estimates from the 5-year American Community Survey
(ACS) covering 2014 – 2018. However, the 2010 Census populations are the
“default” estimate as the breakdown in reporting of racial/ethnic groups in the
ACS is more limited than the Census. To generate the typical Census-based
disparity indexes, each group’s benchmark population only includes non-Hispanic
individuals and all Hispanic individuals are placed in the Hispanic category.
However, the ACS only reports non-Hispanic population totals for Whites,
meaning it is impossible to separate Hispanic individuals from non-White
populations for the purposes of calculating the disparity index. This results
in non-White Hispanic individuals being double counted in the 2018 ACS
disparity index calculations, once in their racial group and once in the
Hispanic group.
Analysis of the 2010 Census data found that 61%
of Missouri’s Hispanic population was White, while 34% would fall into the
“Other” category of this report which consists of those reporting “Other” or
two or more races. This means that the distortion in the benchmark population
for non-White races is limited across the state, though it is possible calculations
for jurisdictions with larger non-White Hispanic populations could be distorted
to a greater degree. The 2018 ACS disparity indexes should be viewed as
supplemental information for analyzing traffic stop disproportions.
Appendices to this report can provide useful analysis to partially alleviate these issues. Appendix A uses statewide rather than local racial/ethnic population portions to generate disparity index calculations, reducing the impact of the residency
issue. Appendix C makes use of the residency question added to the Vehicle Stops Report form for the 2019 reporting cycle to only use traffic stops of residents of an agency’s jurisdiction when generating the disparity index. This allows
an analyst to control for driver residency when examining an agency’s report.
2. Search Rate
A second metric that can be used to assess
traffic enforcement is the “search rate,” or the number of searches divided by
the number of stops (x 100). Searches include both searches of drivers and
searches of the vehicle and property within.
The search rate for all motorists who were
stopped in 2019 was 6.74%. Asians were searched at a rate well below the
statewide average (2.83%), while Blacks (8.86%) and Hispanics (8.30%) were
searched at rates above the average for all motorists who were stopped. Whites
were searched at a rate below the state average at 6.23%. The search rate for
racial/ethnic groups can also be directly compared with one another. Hispanic
drivers were 2.93 times more likely to be searched than Asian drivers
(8.30/2.83).
Limitations
The reasons for conducting a search and the
outcome of the search (i.e. discovery of contraband) should be considered when
making comparisons across groups. Some searches are conducted with the consent
of the driver, or because the officer observed suspected contraband in plain
view, had reasonable suspicion that an individual may possess a weapon (Terry
search), or other reasons. These searches may or may not result in an arrest.
Other searches are conducted incident to
arrest—this means that there is no other reason given for the search other than
arrest. As mentioned above, searches are almost always performed when there is
an outstanding arrest warrant, whether or not contraband may be present.
It is notable that the search rate, along with the other post-stop metrics, are not subject to the benchmarking issues that affect the stop disparity index. The benchmarks for the post-stop metrics are the number of stops reported, so no estimation
is involved in their calculation, making them more reliable metrics for assessing possible bias.
3. Contraband Hit Rate
The third summary metric, the “contraband hit
rate,” reflects the percentage of searches in which contraband is found.
Contraband was found in 35.4% of all searches conducted in 2019. Little
difference exists in the contraband hit rate for searches of White and Black
drivers (35.9% vs. 34.9%). The contraband hit rate for other groups is somewhat
lower. For example, searches of Hispanic drivers resulted in a contraband hit
rate of 28.4%. This means that, on average, searches of White and Black drivers
are more likely than searches of Hispanics to result in the discovery of
contraband.
Limitations
While the contraband hit rate is not subject to benchmark issues, the degree of discretion officers have in their searches should still be considered when examining contraband hit rates. The difference in contraband hit rates among racial/ethnic
groups may result in part from the higher arrest rates for Blacks and Hispanics—if there is an arrest, there will be a search whether or not the arresting officer suspects the subject has contraband. This may deflate the contraband hit
rate since searches are being completed based on procedure rather than suspicion of contraband.
4. Arrest Rate
The “arrest rate” is the fourth summary metric
included in Table 1 that may be useful for assessing bias in traffic
enforcement. Statewide, 4.89% of all traffic stops resulted in an arrest
(74,553/1,524,640).
Approximately
6.21% of the stops of Blacks, 6.34% of the stops of Hispanics, and 4.68% of the
stops of American Indians resulted in arrest, compared with about 4.55% of the
stops of Whites. Asians (2.34%) and people of mixed or unknown race (3.91%) had
lower arrest rates than Whites. The dynamic of Blacks and Hispanics being
arrested at a higher rate likely impacts search and contraband hit rates as
described above.
Limitations
Like the other post-stop metrics, the arrest rate does not suffer from benchmark issues, however, officer discretion should still be considered as described above.
Conclusion
Missouri is a vast and complex state at the crossroads of the country. Each of its communities is unique and presents many dynamics that must be comprehensively evaluated when assessing the possibility of
bias in traffic enforcement. Each agency’s report should be read in conjunction with its entries in each of the appendices and in the context of the dynamics of the community it serves. This report is subject to limitations as described
above and simply aims to provide a starting point for local dialogue.
Appendices
Appendix A: Local Vehicle Stops in Proportion to State Racial Composition
Appendix A presents the traffic stop analysis using the statewide proportions of race and ethnicity, rather than the local proportions of each jurisdiction.
This is valuable as it reduces the impact of the residency benchmark issue on the disparity index of each agency.
Appendix B: Disparity Indexes for 2000-2018 Compared
Appendix B compares the 2019 disparity index to the disparity indexes for 2000 through 2018. For each agency, the disparity index for each race-ethnic group is presented
for 2000-2019. For the state as a whole, the key metrics generally show small changes between 2016 and 2018.
Appendix C: Resident and Non-Resident Stops
Appendix C displays the number and percentage of traffic stops of residents of the jurisdiction where the stop occurred separately from the total number of stops (i.e., stops of
residents + stops of nonresidents). Disparity Indexes for the different population groups are also shown.
Appendix C uses the addition of the driver residency question in the 2018 reporting cycle to provide a disparity index calculated using only traffic stops of drivers who reside in the agency’s jurisdiction. Readers should acknowledge that
a driver’s residency may not accurately reflect where the driver spends the majority of their time. Many Missourians commute from one community to another for work and leisure, so the demographics of drivers on the road may not reflect
those living in the community. By restricting the calculation to only those drivers who reside in a community, the calculation may include drivers who spend the majority of their time in another jurisdiction while excluding those who commute
into the jurisdiction for the majority of the day.
Table 2. Agencies that did not submit reports by March 1, 2020 as required by state law |
---|
Atchison
County Sheriff's Department | Auxvasse
Police Department | Bel-Nor
Police Department* |
Berger
Police Department | Bunker
Police Department | Cardwell
Police Department |
Carrollton
Police Department | Conway
Police Department | Edina
Police Department |
Eminence
Police Department | Everton
Police Department | Fayette
Police Department* |
Frankford
Police Department | Green
Ridge Police Department | Greenwood
Police Department* |
Hayti
Police Department | King
City Police Department | Laddonia
Police Department |
Lowry
City Police Department | Marston
Police Department | Martinsburg
Police Department* |
Rutledge
Police Department | Schuyler
County Sheriff's Department* | Senath
Police Department |
Silex
Police Department | St.
Charles County Parks & Recreation | St.
Mary Police Department |
Sturgeon
Police Department* | Winona
Police Department | |
*Agency did not submit data by the statutory deadline, but did provide data for inclusion in the report.
Table 3. Agencies that reported no stops (many contract out vehicle stops to other agencies) |
---|
Altenburg
Police Department | Armstrong
Police Department | Bellflower
Police Department |
Birch
Tree Police Department | Bunceton
Police Department | Canalou
Police Department |
Centerview
Police Department | Charlack
Police Department | Chilhowee
Police Department |
Clarkson
Valley Police Department | Dudley
Police Department | Emma
Police Department |
Forest
City Police Department | Freeman
Police Department | Jackson
County Drug Task Force |
Jefferson
College Police Department | Marquand
Police Department | Maysville
Police Department |
Mineral
Area College DPS | Miramiguoa
Police Department | Missouri
Department of Revenue |
Parma
Police Department | Risco
Police Department | St.
Louis Community College Police Dept. |
St.
Louis Park Rangers | St.
Peters Ranger Enforcement Div. | Vanduser
Police Department |
Walker
Police Department | Wardell
Police Department | |
[ back to top ]