HomeMy WebLinkAboutBarnum 2019 and 2020 report
Iowa City Police Traffic Study
Brief Summary
2019 and 2020
Prepared by:
Chris Barnum
CR Research Group LC
October, 2021
1
Iowa City Police Traffic Study
In 2019, the City of Iowa City partnered with CR Research Group LC to analyze the
Iowa City Police Department’s traffic stop activity. The review focused on evaluating stops
made by the department between January 1st 2019 and December 31st 2020. These analyses
evaluated two broad categories of discretionary police conduct: (i) racial disparity in vehicle
stops—expressed as racial differences in the likelihood of being stopped by the police and (ii)
dissimilarities across racial demographics in the outcome or disposition of a stop.
To evaluate the likelihood of being stopped, our research team utilized driver-population
benchmarks fashioned from roadside observations and census data. A benchmark should be
thought of as the racial proportion of drivers on the roads in a given location. At its best, the
benchmark is a standard that can be used to judge the percentage of drivers that should be
stopped by the police when no bias is occurring. In Iowa City, the population characteristics of
the city were divided up into one-square-mile units called observation zones’ (see below).
Figure 1. City of Iowa City Observation Zones.
Once the boundaries of the observation zones were determined, roadside surveyors were
deployed to monitor traffic at several locales within selected zones. The observers focused their
efforts on surveying traffic in areas where the ICPD tended to make most of their stops. The
observers watched traffic at various times of the day ranging from 7:00 am until 2:00 am. To
date, surveyors have logged more than 110,000 observations in several waves from locations
across the city. Please see table A1 in appendix-I for a listing of the number of traffic
2
observations by year and a figure showing the zones where traffic observations were
concentrated.
The process of comparing police data to benchmarks is straight forward. It centers on
identifying differences between the demographic percentages from ICPD traffic stop data and
benchmark information. Any positive difference between benchmark values and police data
signifies disproportionality or an over representation of nonwhite drivers in the data. The
benchmark used for this report accounts for the potential of masked disproportionality.1
Although, disproportionality can indicate bias or discrimination, it does not necessarily signify
bias. It is possible for disproportionality to occur for a number of reasons, including differences
between racial groups in driving behavior, vehicle condition, driver-license status and so forth.
Our methodology makes it possible to track disproportionality by area of town, by time
of day, by duty assignment and by individual officer. While this method serves as a useful tool in
assessing disproportionality, please keep in mind that the method is only an estimate of
disproportionality. As noted, the analyses are predicated on benchmark information and the
benchmarks are formed from samples of the drivers on the roads in a given area and time.
Consequently, like any sample, a benchmark may be associated with a degree of uncertainty or
indeterminacy. This means that numerical estimates of disproportionality are likely associated
with some error and the true population parameter may be larger or smaller than the estimate.2 In
what follows, we present a summary measure of disproportionality. This index can take on both
positive and negative values, with zero signifying no disproportionality. However, given
sampling error, smaller index values near zero do not necessarily indicate disproportionality
because such values could be due to chance alone.
Analyses
The charts below give the number of ICPD traffic stops in 2019 and 2020 for each square
mile zone after removing data containing missing values. The information indicates that for both
years, most stops occurred in the downtown area of the city (zone 21) followed by Broadway-
Wetherby (zone 29) and surrounding areas (zones 28, 30 and 13). In point of fact, for both years,
roughly forty percent of all ICPD traffic stops occurred in zone 21, although the percentage
tended to be higher during nighttime hours than during the day.3 The analyses also show that the
ICPD made 7,614 fewer traffic stops in 2020 when compared to 2019, and this is likely due to
COVID-19. As noted, each observation zone consists of a square mile area rather than a specific
intersection, table A2 in appendix-I gives the approximate street boundaries for the two busiest
zones, 21 and zone 29.
1 Previous analyses suggest that certain non-white racial categories tend to be stopped at lower rates than their
actual percentages in the driving population. When this is the case, grouping all people of color together as a
single unit could mask disproportionality. For this study, we grouped white drivers and Asian drivers together to
form our benchmarks. According to the 2019 US Census American Community Survey, Asians comprised roughly
7.3% of the population in Iowa City. Analyses of ICPD traffic stop data show that in 2019 Asian drivers comprised
roughly 5.67% of all drivers stopped; and in 2020 Asian drivers comprised about 4.6% of all drivers stopped by the
ICPD.
2 Sources of variation and sampling error include, variability of the traffic flow itself within observation zones,
variability between roadside surveyors, variability of racial proportions of residents within observation zones,
choice of locations to record traffic characteristics within a zone, and variability associated with assigning stops
made on observation zone borders.
3 Please refer to the supplemental charts in appendix-II for the percentage of stops by zone and by time-of-day.
3
Disproportionality Index Values
Table 1. below gives information for the summary disparity index broken out in three ways, for:
(i) all officers, (ii) officers working days and (iii) officers working nights. The index gives an
estimate of disproportionality using a weighted average. The index is computed by summing the
weighted difference between percentage of police stops involving nonwhite drivers for a given
observation zone and corresponding benchmark values. Weights consist of the number of stops
made in each zone.
2019 Number of stops = 14,111
2020 Number of Stops = 6497
4
Table 1. Disproportionality Index Values for 2019 and 2020
Date Level of Analysis
Department Days Nights
2019 0.089 0.063 0.100
2020 0.073 0.028 0.108
For 2019, the disparity index for the department equaled 0.089, signifying that on
average, across all observation zones, the proportion of traffic stops involving drivers who were
identified as people of color was about nine percentage points higher than corresponding
benchmark proportions. The information also suggests that levels of disproportionality were
higher during nighttime hours than during the day with values at night about four percentage
points higher than for days.4
The information for 2020 shows that the overall level of disproportionality for the
department decreased when compared to 2019. In 2020, the disparity index for the department
equaled 0.073, which was about one-and-a-half percentage points lower than in 2019. As in
2019, the 2020 information suggests that level of disproportionality was higher during nighttime
hours than during the day, with nighttime disproportionality about eight percentage points higher
than days. For both years the highest levels of disproportionality tended to be clustered in
specific zones, especially at night. In general, the largest concentrations of disproportionality
were centered in the downtown area (zone 21) and areas outside our observation areas. Also, for
both years ICPD officers made more stops at night than during daytime hours. In 2019, officers
working at night made 9982 of the 14,111 total stops and in 2020, night officers made 3,673 of
the 6,497 total stops. Table 2. below shows that departmental index values since 2015 have
generally been stable with no discernable trend (an index value is not available for 2017).
Table 2. Index Values by Year.
The information in table 2 probably gives the upper bounds for estimates of
disproportionality especially for the most recent years. As noted, when establishing benchmarks,
we concentrated roadside traffic observations in areas of town where the ICPD tended to make
most of their stops. In other areas we used information from the 2010 Census report to estimate
benchmarks. The census information suggests that in 2010, about 10% of the population in Iowa
City did not identify as white or Asian. Accordingly, we used that assessment to estimate a 10%
benchmark in unobserved areas. However, the racial demographics of Iowa City have
undoubtedly changed since 2010. In fact, the 2019 Census Bureau’s, American Community
4 We consider stops made between 7:00am and 6:59pm to be daytime stops. All others are considered nighttime
stops.
Year Index
2015 0.05
2016 0.08
2018 0.07
2019 0.09
2020 0.07
5
Survey, suggests that in 2019, roughly 14% of the population in Iowa City did not identify as
white or Asian. For continuity, and for comparability with earlier reports, we have continued to
report index values based on 2010 census benchmark information. However, given that actual
percentage of minority drivers may be higher than the 10% benchmarks indicate, the reported
index results may be marginally high. For instance, we also computed the 2020 department index
using 2019 ACS information for benchmarks (14%) instead of 2010 census information. The
resulting index value using this modified benchmark was lower, equaling 0.067 instead of the
original 0.073. Additionally, although the 2020 US Census Bureau information for racial
demographic percentages for Iowa City was not available for this report, it seems likely that the
percentage of nonwhite residents will increase from 2019 estimates. If this is the case, then the
index value based on the 2020 benchmark would be lower than the 0.067 estimate given above.
Consequently, when interpreting index scores through time readers should keep in mind that
recent index values are likely on the upper range of estimates of disproportionality.
It is also important to keep in mind that the benchmarks should be interpreted as an
estimate of the percentage of nonwhite drivers on the roadways and that differences between
benchmark values and stop percentages should be understood as approximations of aggregate
disproportionality in the population.
Officer Level Analysis:
We calculated a disparity index for each officer making one hundred stops or more
during 2019 and 2020.5 The index consists of two ratios, and is computed by comparing the
fraction of stops involving minority drivers to corresponding benchmarks divided by the
proportion of stops involving other drivers to their corresponding benchmarks. These values are
weighted by the number of stops and summed across all zones.6 Higher absolute values suggest
more disproportionality.
The index can take on values from positive to negative infinity, with values in the range
between zero and one indicating little or no disproportionality. As a rule, these charts are most
useful in a qualitative way, because they are an expedient internal benchmarking instrument for
comparing officers to one another. The charts facilitate identifying officers with comparatively
high and dissimilar index values. Such officers would show up as a solitary dot, located above
the blue dashed line and on the extreme right side of a chart. It is important to use caution when
interpreting index values calculated from a relatively low number of stops (especially, fewer than
100 stops). Index calculations predicated on comparatively few stops can be quite unstable and
change significantly with the addition or subtraction of only a couple of stops. The stability of
the index increases as the number of stops increase. Additionally, we suggest police managers
should use additional internal benchmarking7 techniques to supplement interpretations of index
5 Previous charts showed information for officers making twenty-five stops or more. We changed to this threshold
for this report due to the comparatively large number of officers with fewer than one-hundred stops in 2020.
6 Initial index values can range solely between zero and positive infinity. However, in computing reported index
scores, the values between zero and one in each zone are converted to their negative reciprocal and all scores are
then weighted and summed. Please note there are at least two sources of indeterminacy in computing index
values. The first is the previously mentioned potential sampling error associated with benchmark estimates. The
second source of indeterminacy is that the index is undefined when the denominator equals zero. This generally
occurs when very few stops are made in a given zone. In these circumstances the index is made to generate a unit
value.
7 Samuel Walker, 2003. Internal benchmarking for traffic stop data: an early intervention system approach.
Retrieved from: www.policeaccountability.org
6
results, especially for any officers identified with comparatively high index values. For instance,
managers should contrast such officers with similarly situated officers, including those who work
the same shifts, beats, duty assignments, special projects and so forth in order to gain additional
insight into index interpretations. Finally, it is important to recognize that an individual index
value reflects a single snapshot in time. And given the indeterminacy associated with computing
the index, it is important to interpret outcomes by looking for trends through time.
The two figures below give the disparity index values and number of stops for officers
making at least 100 traffic stops in 2019 and 2020. For each graph, a dot represents an officer,
the blue horizontal line indicates 100 stops made, the thick red dashed line shows the median
disparity index value for all officers making at least 100 stops and the thin red dashed line gives
the index 90th percentile value for all officers making 100 stops. The information suggests that
officers’ index values are generally clustered together for both years with no noteworthy outliers.
7
A comparison of these charts with previous years suggests that the officers with the
highest index values in 2019 and 2020, had somewhat lower index values than officers with the
highest values in previous years, and that in general, there are no distinct outliers in the recent
charts when compared to earlier years. Please see appendix-III for charts from previous years.
Stop Outcomes Results
We used an examination of stop outcomes to assess disproportionality in citations, arrests
and consent searches. As the name implies, a stop outcome gives information about the
consequence of a stop. An example of an outcome is whether or not a driver received a ticket as
a result of the stop. In what follows we measure disproportionality using a statistic called an
odds ratio. This estimator is a measure of effect size and association. It is useful when comparing
two distinct groups and summarizes the odds of something happening to one group to the odds of
it happening to another group. An odds ratio value greater than one indicates an increased
occurrence of an outcome for a nonwhite driver. Analyses of odds ratios are an excellent way to
identify trends in the data. Table 3 below gives the odds ratios for stop outcomes for 2011 –
2020.
Table 3. Department Outcomes and Univariate Odds Ratios by Year
Outcome
2011 2012 2013 2014 2015 2016 2017 2018 2019 2020
Citations 1.4 1.4 1.6 1.5 1.3 1.4 1.07 1.0 -1.02 1.55
Arrests 3.2 2.5 2.3 2.1 1.9 1.5 1.82 1.98 2.32 2.08
Requests 3.9 2.4 1.9 1.5 1.9 2.1 ---- --- ---- ---
The information in Table 3 shows that the odds ratios for each outcome have remained
generally stable in recent years. In 2020, while the odds ratios for citations increased when
compared to 2019 values, the odds ratios for arrests decreased. It is noteworthy that in both
years, the ICPD tended to issue citations to less than 15% of drivers stopped.8 In both 2019 and
2020, ICPD officers initiated very few consent searches. Consequently, we could not reliably
calculate odds ratios for this outcome.9
Looking more closely at arrests, it is important to keep in mind that a large majority of
the arrests made in both 2020 and 2019 were for nondiscretionary charges. These are offenses
that due to state law or departmental policy, leave officers with very little or no choice in
deciding whether or not to make an arrest. Officers are in essence required to arrest, and would
in fact, be subject to departmental discipline if they chose not to arrest. These types of charges
include offenses like bench warrants, driving while barred and operating while intoxicated.
Analyses show that in the overwhelming majority instances where an arrest was made, officers
had little choice in the matter. Please see appendix-IV for tables showing information regarding
the number and percentage of nondiscretionary arrests as well as arrest information by
benchmark.
8 In 2019, ICPD officers issued citations to 13.87% of drivers stopped; in 2020 ICPD officers issued citations to
8.30% of drivers stopped.
9 In 2019, ICPD officers initiated five total consent searches, one of these involved a minority driver. In 2020,
minority drivers accounted for three of the eight consent searches requested. The reduction in number of requests
likely stems from ICPD training that occurred around 2017 focusing on how to properly code search requests into
the reporting system. In previous years, the ICPD reports that some information may have been miscoded.
8
Conclusions
This study examined the traffic stop behavior of the Iowa City Police Department using
data from 2019 and 2020. The investigation focused on two broad categories of police conduct,
racial disproportionality in vehicle stops (at both the agency level and officer level) and
disproportionality in the outcome or disposition of a stop. Findings from the examination of
disproportionality in vehicle stops suggests steady or decreasing amounts of racial
disproportionality in traffic stops and on average, the agency-level index for drivers who were
identified as not white or not Asian decreased in 2020 from 2019. Additionally, it seems likely
that analyses based on projected 2020 US Census benchmarks will be lower than current
estimates. Consequently, it is likely that the indices for the most recent years probably give the
upper bounds for estimates of disproportionality. Together, these results provide little or no
evidence of a rise in disproportionality in stops for the agency during the study period and it
seems likely disproportionality is decreasing. Analyses of officer level data indicated that
although officers’ index values were generally clustered together with similar index values for
the most recent years, there were earlier occasions that showed officers with distinctly higher
disparity index values. The findings from all years however, do not suggest that these officers
continued markedly higher disproportionality levels into 2019 or 2020. Finally, the results for the
analyses of stop outcomes generally indicates comparatively low levels of disproportionality in
stop outcomes for citations, but higher levels of disproportionality in arrests. The
disproportionality in arrests however, does not appear to be trending higher over the last three
years. And it is important to note that almost all arrests in recent years were made for
nondiscretionary offenses, meaning the officer had little or no choice in deciding to make an
arrest.
Some comments and recommendations for future work. First, if possible, the ICPD
should include exact latitude and longitude information in the traffic stop data sets. With the
existing data, it is not possible to locate the exact position of stop locations. Instead, stops are
grouped into one square mile areas. This lack of precision increases potential error in estimating
index values. However, if it were possible to locate stops at a more granular level, it would then
be possible to make much more precise estimates of disproportionality. Second, the results show
that a sizeable majority of stops made by Iowa City officers were for offenses that were not
issued a traffic citation. Police managers should assess and evaluate the reasonableness of the
practice of stopping large numbers of vehicles for offenses that officers deem deserving of only a
warning. Third, if possible future data should include additional information concerning
occupants of the vehicle, specifically information regarding whether driver or occupants were
asked to step out of the vehicle, and a record of warrant and registration requests (for both
vehicle and occupants). Finally, police managers should use the information from
disproportionality analyses to look closely at disproportionality found at the officer level. We
suggest that supervisors use additional internal benchmarking techniques to compare officers to
similarly situated officers (e. g., other officers working the same time, duty assignment, beat and
so forth) to determine if these structural factors may account for some or all of the observed
disproportionality.
9
Appendix I
Table A1.
Traffic Observations by Year
Year Observations
2018 54,218
2015 27,032
2007-12 28,951
Total 110,201
Table A2.
Selected Observation Zone Approximate Street Boundaries
Zone 21 Zone 29
North E. Market Street Kirkwood Avenue
South Kirkwood Avenue Wetherby Park
East Summit Street Taylor Drive
West Iowa River Iowa River
Figure A1. Areas where traffic observations were concentrated
10
Appendix II
Supplemental Stop Charts
11
12
Appendix III
Officer Index Charts Years 2015 - 2020
Note: Outliers are indicated with an arrow. Officers showing up below the blue horizontal line
made fewer than 100 stops. Indexes associated with a limited number of stops should be
considered unreliable.
2020 2019
2018 2017
2016 2015
050010001500Number of Stops0 2 4 6
Disparity Index 050010001500stops0 2 4 6index
13
Appendix IV
Stop and Outcome Information
Table A3
2019 Raw Information*
Race No. Arrests Non-Discretionary Percent Non-Dis. No. Stops
White 448 417 93.1 9759
Asian 32 31 96.8 821
Black 251 230 91.6 2583
Hispanic 94 88 93.6 845
Native 2 2 100 26
Other 4 4 100 91
Unknown 8 8 100 355
Totals 839 780 92.9 14480
* Totals for stops analyzing stop index values shown in main part of document (N= 14111) exclude rows
with missing race information (355), stop location (8) and outcome information (6).
Table A4
2019 Arrest Numbers by Benchmark*
Race No Arrest Yes Arrest Total
Minority 3190 351 3541
W&A 10096 479 10575
Total 13286 830 14116
* Totals exclude missing and unknown values, including 355 stops with unknown race information, eight
stops with missing stop zone information and one missing value for arrest type.
See next page
14
Table A5
2020 Raw Stop Information*
Race No. Arrests Non-Discretionary Percent Non-Dis. No. Stops
White 254 243 95.6 4718
Asian 10 9 90 298
Black 107 103 96.2 1123
Hispanic 44 42 95.5 328
Native 3 3 100 16
Other 0 0 --- 15
Unknown 2 2 100 190
Totals 420 402 95.7 6688
* Totals for stops analyzing stop index values in main part of document (N = 6497) exclude rows with
missing race information (190) and outcome information (1).
Table A6
2020 Arrest Numbers by Benchmark*
Race No Arrest Yes Arrest Total
Minority 1327 154 1481
W&A 4752 264 5016
Total 6079 418 6497
* Totals exclude missing and unknown values including 190 stops with unknown race information and
one missing value for outcome type.