See pages 37 to 39
REPORT TO THE WASHINGTON STATE PATROL
Relating to:
National Highway Traffic Safety Administration (NHTSA) Grant-Funded Study on
Racial Profiling Phenomena in Washington State
OGRD # 107828

Results of the Monitoring of WSP Traffic Stops for Biased
Policing: Analysis of WSP Stop, Citation, Search
and Use of Force Data
and
Preliminary Results of the Use of Observational Studies for
Denominator Assessment
Division of Governmental Studies and Services
Washington State University
Research Team
Nicholas P. Lovrich, Ph.D.
Michael J. Gaffney, J.D.
Clayton C. Mosher, Ph.D.
Travis C. Pratt, Ph.D.
Mitchell J. Pickerill, J.D., Ph.D.

Support Staff:
Analysis and Report Preparation
Yu-Sheng (Linus) Lin, A.B.D., Data Management & Statistical Analysis
Michael J. Erp, M.A., Field Observation Oversight and Report Prep

Project Support
Ruth Self, Office Coordination

 Christina Sanders, M.P.A., Research Coordinator
Julie Lusby, Budget and Finance Coordinator
Heidi Lee, Travel Coordinator
September 15, 2007

Table of Contents
Executive Summary

2

Analysis of Traffic Stops for Evidence of Biased Policing

3-17

Analysis of Citations for Evidence of Biased Policing

17-35

Analysis of Searches for Evidence of Biased Policing

35-44

Analysis of Use of Force for Evidence of Biased Policing

44-49

Preliminary Results of the Use of Observational Studies for
Denominator Assessment

50-58

Executive Summary
This portion of the final report prepared under the auspices of the National
Highway Traffic Safety Administration (NHTSA) Grant-Funded Study on Racial Profiling
Phenomena in Washington State [OGRD # 107828] sets forth findings derived from the
independent monitoring of traffic stop data collected by the WSP. This report sets forth
the results of an analysis of traffic stops, traffic citations, searches and use of force for
evidence of biased policing. Our analysis of agency data is carried out both at the
statewide and individual Autonomous Patrol Area (APA) levels. Our analysis indicates
very few instances of noteworthy minority/non-minority disparities in the use of police
discretion by the officers of the Washington State Patrol. Most importantly, there is no
evidence of a systematic practice of racial profiling in either who is stopped, who is
issued a citation, who is searched, and to whom force is applied by WSP officers. In
addition to these substantive findings, this report also sets forth findings derived from a
testing of the utility of racial coded traffic stop data as a “denominator” for racial

2

 profiling assessments by means of three observational studies conducted with digital
photography. Those results indicate that accident data are likely to represent a reliable
and cost-effective indicator of driver population demographics, making the monitoring of
racial profiling an affordable practice in nearly all police jurisdictions.

The Analysis of Traffic Stops for Evidence of Biased Policing:

Self-

Initiated Contacts
Table 1 presents data on all (self-initiated) traffic stops by the Washington State
Patrol for the November, 1, 2005 to September 30, 2006 period for each of the 34
autonomous patrol areas (APAs). Statewide, 83.1% of those contacted by the WSP were
White; 3.7% were African-American, 0.6% Native-American, 3.6% Asian/Pacific
Islanders, 0.9% East Indian, and 7.8% Hispanic. Comparisons of these data to 2005 U.S.
Census Bureau data on the racial/ethnic composition of Washington State indicate that
Whites are slightly under-represented in WSP traffic stops (Whites comprise 85% of
Washington State’s population); Blacks are slightly over-represented (3.5% of
Washington State’s population); Native-Americans are under-represented (1.7% of
Washington State’s population); Asian/Pacific Islanders are under-represented (6.9% of
Washington State’s population) and Hispanics are slightly under-represented (8.8% of
Washington State’s population) (the Census Bureau does not provide data on the
percentage of East Indians).
(Table 1 on the following page)
Census data are not ideal benchmarks in analyses of traffic stop data, as there are
likely to be differences in driving patterns and the types/conditions of vehicles across
racial groups that may have an impact on who is contacted (see Lorie A. Fridell, By the
Numbers: A Guide for Analyzing Race Data from Vehicle Stops, Washington, DC:
3

 Police Executive Research Forum, 2004). In addition, particularly with respect to the
Hispanic population in Washington State, U.S. Census data may underestimate the total

Table 1: WSP Trooper Self-Initiated Contacts (%)
Data for Nov. 2005 - Sept. 2006
APA
2
3
4
5
6
7
8
11
12
13
14
15
16
19
20
21
22
23
24
25
26
27
28
30
31
32
33
34
35
36
37
38
39
40

White
74.3
82.1
85.3
68.3
66.1
74.6
90.3
60.1
49.5
73.7
84.5
97.0
88.1
93.9
93.6
86.2
88.9
87.8
85.4
80.5
85.6
82.0
77.5
81.0
83.7
88.1
78.6
90.0
92.6
86.2
88.4
89.5
94.1
94.6

Black
11.7
7.4
4.5
10.9
12.7
5.5
1.7
1.9
1.2
2.2
1.0
0.4
2.4
2.2
1.8
3.4
0.8
3.1
2.7
0.5
2.7
0.4
2.1
2.3
2.1
4.7
4.3
1.8
1.3
5.2
2.1
1.6
0.8
0.5

Native Amer.
0.4
0.2
0.6
0.2
0.3
0.1
1.0
5.8
1.8
0.2
0.5
1.5
0.5
0.9
0.3
0.2
1.1
0.1
0.2
0.2
0.5
4.1
0.3
1.6
0.4
0.3
0.4
0.1
1.8
0.4
0.8
1.5
0.2
0.3

4

Asian
6.4
4.0
4.1
10.2
9.0
8.6
2.6
1.4
1.0
1.5
0.8
0.3
2.5
1.1
2.5
3.2
0.9
3.1
3.9
1.1
3.0
0.5
2.0
7.2
4.4
3.0
7.0
2.6
2.1
4.0
2.6
1.6
1.6
0.9

E. Indian
0.6
0.4
0.6
2.2
1.7
2.5
0.4
0.4
0.3
0.2
0.3
0.1
0.7
0.1
0.3
0.8
0.3
0.8
1.2
0.4
0.7
0.2
0.6
3.3
1.4
0.2
2.2
0.5
0.4
0.2
0.3
0.1
0.2
0.2

Hispanic
6.0
5.5
4.4
7.5
9.3
8.1
3.6
30.2
45.8
21.9
12.9
0.6
5.7
1.6
1.3
5.7
7.8
4.6
6.5
17.2
7.2
12.8
17.2
4.1
7.8
3.7
7.2
4.9
1.8
3.7
5.3
5.5
3.0
3.3

N
15,404
13,895
16,060
14,656
19,549
21,918
3,504
11,564
5,423
18,401
10,891
8,079
10,928
23,481
8,010
20,553
7,028
11,123
11,425
18,192
17,514
7,977
15,753
11,405
10,693
8,472
31,618
16,837
17,576
26,421
14,200
4,616
8,758
6,139

  

Statewide
569,862

83.1

3.7

0.6

3.6

0.9

7.8

resident population due to the presence of migrant workers and undocumented
immigrants. It is also important to note that certain areas of the state (particularly the
Interstate -5 corridor running from the Canadian border to the Oregon border) patrolled
by the WSP have a high proportion of out-of-state drivers, and it is probable that these
drivers are more likely to be members of racial minority groups than resident in-state
drivers. Finally, census data are particularly problematic to use as benchmarks when
analyzing data from smaller geographic units, such as autonomous patrol areas. As such,
our analyses utilize four alternative benchmarks which we have argued in prior reports
represent a comprehensive source of “denominator estimates” (see Nicholas Lovrich,
Michael Gaffney, Clay Mosher, Mitchell Pickerill, and Michael Smith, Washington State
Patrol Traffic Stop Data Analysis Project Report, June, 2003) contacts initiated as a
result of “calls for service” and vehicle assists, contacts initiated as a result of radar
patrols; WSP contacts initiated in responding to collisions; and daytime traffic stops.
In these analyses, we adopt the criterion used in several other studies of racial
profiling that differences are not substantively significant as long as the percentage of
those contacted in any particular racial group is not more than five percentage points
greater than the percentage of the group in the benchmark comparison 1 (see Joyce
1 Alternative measures of disparity include the “ratio of disparity,” “relative differences,”
and the “disparity index,” (Fridell, 2004) or what Lamberth (see Lamberth et al., Ann
Arbor Police Department Traffic Stop Data Collection Methods and Analysis Study,
Report submitted to the Ann Arbor Police Department by Lamberth Consulting, 2004)
refers to as “odds ratios.” The latter measure is calculated by dividing the percentage of
drivers in a particular racial group who are stopped by their percentage in the benchmark
population. As Fridell (2004) notes, when the percentage in a particular minority group
in both the contacted driver population and the benchmark population is low, the disparity
index (and the two alternative measures of disparity) can be misleadingly high. Although
there are certain APAs in which the proportion of minorities (particularly Hispanics)
contacted is relatively high, at the statewide level no racial minority group represents
more than 7.1% of those contacted by the WSP. Thus, in order to maintain consistency in
the reporting of our results, and in order to avoid the presentation of potentially
6

 McMahon, Joel Gardner, Ronald Davis, and Amanda Kraus, How to Collect and Analyze
Racial Profiling Data: Your Reputation Depends On It! Final Project Report for
Racial Profiling Data Collection and Analysis. Washington, DC: US Government
Printing Office, 2002).
Calls for Service and Self-Initiated Vehicle Assists
The WSP data include a separate code for contacts initiated as a result of calls for
service and vehicle assists. This particular benchmark can be considered a “blind” type
of benchmark because it is highly unlikely that WSP Troopers would know the race of the
individual being assisted in the vast majority of such citizen contacts. Table 2 displays
findings on the percent of drivers contacted by WSP Troopers as a result of calls for
service and vehicle assists by race and APA (due to reliability concerns, analyses were
restricted to APAs where there were a minimum of 20 such WSP Trooper contacts over
the November 1, 2005 to September 30, 2006 period). The cell entries in Table 3
represent the figure obtained after subtracting the percentage of individuals contacted as a
result of calls for service and vehicle assists from the percentage of all self-initiated
contacts in each APA.
(Tables 2 and 3 on the following two pages)
The findings set forth in Table 3 indicate that there are no APAs in which the
percentage of Blacks, Native-Americans, or East Indians contacted as a result of selfinitiated WSP activity is more than five percentage points greater than those contacted as
a result of calls for service and vehicle assists. For Asians/Pacific Islanders, there is one
Table 2: Calls for Service and Vehicle Assists
misleading findings, our measure of disparity subtracts the percentage of those in each
racial group contacted from their percentage in the various benchmarks.
7

 Data for Nov. 2005 - Sept. 2006
APA
2
3
4
5
6
7
8
11
12
13
14
15
16
19
20
21
22
23
24
25
26
27
28
30
31
32
33
34
35
36
37
38
39
40
Statewide

White
77.3
80.1
88.2
72.6
65.3
78.4
76.7
45.5
44.8
68.0
55.2
—
78.6
91.5
92.5
82.7
—
89.7
75.0
72.4
78.8
—
60.0
89.6
79.6
—
79.3
92.6
84.6
86.2
79.5
89.5
—
—
79.3

Black
11.1
7.9
3.5
10.0
14.7
2.7
3.3
5.0
3.4
4.0
6.9
—
2.9
3.0
1.9
4.6
—
4.4
8.9
0.6
3.0
—
14.3
2.1
2.2
—
3.8
0.4
2.6
5.1
2.3
2.6
—
—
5.6

Native Amer.
—
0.8
—
—
1.0
0.5
3.3
11.9
6.9
—
6.9
—
—
1.3
—
—
—
1.5
0.8
0.6
2.0
—
5.7
—
—
—
0.4
—
7.7
0.7
2.3
5.3
—
—
0.9

8

Asian
6.0
4.9
3.5
9.1
7.8
7.8
3.3
3.0
—
—
—
—
3.9
0.4
3.8
3.6
—
—
3.2
0.6
2.0
—
2.9
2.1
1.1
—
7.1
1.2
2.6
3.4
2.3
—
—
—
4.8

E. Indian
0.7
—
0.6
1.4
2.0
1.6
10.0
1.0
—
—
—
—
—
0.4
—
0.7
—
1.5
1.6
2.9
—
—
—
—
6.5
—
1.7
0.8
—
0.3
—
—
—
—
1.1

Hispanic N
4.8
415
6.3
493
4.1
170
6.6
351
8.1
409
8.6
370
10.0
30
33.7
101
44.8
29
28.0
50
31.0
29
—
4
14.6
103
2.1
235
1.9
53
8.5
307
—
14
2.9
68
10.5
124
23.0
174
14.1
99
—
18
17.1
35
6.3
48
10.8
93
—
10
7.3
1,133
4.5
242
2.6
39
4.2
731
13.6
44
2.6
76
—
16
—
6
8.0
6,119

 Table 3: Self-Initiated Contacts Minus Contacts Via
Calls for Service and Vehicle Assists (%)
Data for Nov. 2005 - Sept. 2006
APA
2
3
4
5
6
7
8
11
12
13
14
15
16
19
20
21
22
23
24
25
26
27
28
30
31
32
33
34
35
36
37
38
39
40
Statewide

White
-3.0
+2.0
-2.9
-4.3
+0.8
-3.8
+13.6
+14.6
+4.7
+5.7
+29.3
—
+9.5
+2.4
+1.1
+3.5
—
-1.9
+10.4
+8.1
+6.8
—
+17.5
-8.6
+4.1
—
-0.7
-2.6
+8.0
0.0
+8.9
0.0
—
—

Black
+0.6
-0.5
+1.0
+0.9
-2.0
+2.8
-1.6
-4.1
-2.2
-1.8
-5.9
—
-0.5
-0.8
+0.1
-1.2
—
-1.1
-6.2
-0.1
-0.3
—
-12.2
+0.2
+0.1
—
+0.5
+1.4
-1.3
+0.1
-0.2
-1.0
—
—

Native Amer.
+0.4
-0.6
+0.6
+0.2
-0.7
-0.4
-2.3
-6.1
-5.1
+0.2
-6.4
—
+0.5
-0.4
+0.3
+0.2
—
-1.4
-0.6
-0.4
-1.5
—
-5.4
+1.6
+0.4
—
0.0
+0.1
-5.9
-0.3
-1.5
-3.8
—
—

Asian
+0.4
-0.9
+0.6
+1.1
+1.2
+0.8
-0.7
-1.6
+1.0
+1.5
+0.8
—
-1.4
+0.7
-0.7
-0.4
—
+3.1
+0.7
+0.5
+1.0
—
-0.9
+5.1
+3.3
—
-0.1
+1.4
-0.5
+0.6
+0.3
+1.6
—
—

E. Indian
-0.1
+0.4
0.0
+1.6
+0.3
+0.9
-9.6
-0.7
+0.3
+0.2
+0.1
—
+0.1
-0.3
+0.3
+0.1
—
-0.7
-0.4
-2.5
+0.7
—
+0.6
+3.3
-5.1
—
-1.2
-0.3
+0.4
-0.1
+0.3
+0.1
—
—

Hispanic
+1.2
-0.8
+0.3
+0.9
+1.2
-0.5
-6.4
+12.7
+ 1.0
-6.1
-18.1
—
-8.9
-0.5
-0.6
-2.8
—
+3.6
+6.7
-5.8
-6.9
—
+0.1
-2.2
-3.0
—
-3.4
+0.4
-0.8
-0.5
-8.3
+3.1
—
—

+4.2

-1.9

-0.3

-1.2

-0.2

-0.2

9

 APA (APA 30 - Bellingham) for which the difference is greater than five percent, and for
Hispanics, there are two APAs for which the difference is greater than five percent (APA
11 - Yakima, and APA 24, Chehalis).
Radar Patrols
A second benchmark available for analysis is the comparison of traffic stop data
for drivers who have been contacted as a result of being identified as speeding via radar
with all other stops. This particular benchmark statistic constitutes a measure of both
driving quantity and driving quality, and has the important additional advantage of being
a “blind” count – that is to say, WSP Troopers operating radar units seldom if ever can
determine the race of motorists identified as speeders by this traffic safety enforcement
technique.
The figures displayed in Table 4 present findings on the percent of drivers
contacted by the WSP as a result of radar displayed by race and APA, and the figures
presented in Table 5 subtract the percentage of contacted via radar (by race) from the
percentage contacted by the WSP as a result of all other self-initiated contacts. Adhering
to the above-mentioned standard of differences of greater than five percent being
substantively significant, Table 5 reveals that there is not a single APA in which Blacks,
Native-Americans, Asian/Pacific Islanders, or East Indians are over-represented in
contacts initiated as a result of radar patrols compared with all other self-initiated
contacts. Hispanics are over-represented in one APA (APA 12 - Sunnyside).

(Tables 4 and 5 on the following two pages)

10

 Table 4: Contacts Via Radar Patrols (%)
Data for Nov. 2005 - Sept. 2006
APA
2
3
4
5
6
7
8
11
12
13
14
15
16
19
20
21
22
23
24
25
26
27
28
30
31
32
33
34
35
36
37
38
39
40
Statewide
271,168

White
76.4
83.2
84.4
69.7
69.6
76.1
90.5
65.5
55.0
74.3
84.7
97.4
87.8
94.6
93.8
86.3
90.1
85.7
85.1
83.4
85.9
85.3
80.5
78.4
83.5
89.5
78.6
90.2
92.1
86.6
88.2
90.4
94.4
94.4

Black
9.7
7.0
4.7
8.1
10.9
4.4
1.5
2.0
1.5
2.1
1.0
0.4
2.5
1.7
1.8
2.9
0.7
3.9
2.9
0.6
2.7
0.4
2.3
2.0
2.4
4.1
3.8
1.8
1.2
5.0
2.8
1.7
0.8
0.7

Native Amer.
0.2
0.2
0.5
0.2
0.2
0.1
0.8
3.8
1.4
0.2
0.5
1.2
0.6
1.0
0.3
0.2
0.7
0.1
0.2
0.3
0.5
2.9
0.3
1.0
0.3
0.2
0.3
0.1
1.7
0.4
0.6
1.2
0.2
0.3

Asian
6.8
4.0
4.3
11.3
7.9
8.2
3.2
1.8
1.2
1.5
0.9
0.4
2.7
1.0
2.4
3.4
0.9
4.0
4.4
1.4
3.4
0.6
2.4
9.9
6.4
2.8
8.4
2.6
2.5
4.1
3.0
1.9
1.7
1.0

E. Indian
0.5
0.4
0.8
2.7
1.6
2.6
0.4
0.5
0.3
0.2
0.4
0.1
0.7
0.1
0.3
1.0
0.2
1.0
1.3
0.4
0.7
0.2
0.8
4.2
1.8
0.2
2.5
0.6
0.5
0.2
0.4
0.3
0.3
0.2

Hispanic
5.8
4.7
4.5
7.2
8.5
7.9
3.3
26.2
39.9
21.5
12.5
0.5
5.7
1.5
1.2
5.7
7.4
4.5
5.9
14.0
6.5
10.5
13.4
3.8
5.3
3.0
6.1
4.6
1.8
3.5
4.4
4.5
2.6
3.4

84.8

2.9

0.5

3.6

0.9

6.8

11

N
3,313
4,092
5,567
3,644
5,329
7,691
2,190
5,651
2,806
9,283
7,594
3,702
7,156
9,482
5,054
7,745
5,027
4,994
7,810
9,866
10,096
5,599
8,949
6,272
5,667
3,646
14,392
8,516
10,561
13,207
8,241
1,975
6,664
4,000

 Table 5: Self-Initiated Contacts Minus Contacts Via Radar Patrols
Data for Nov. 2005 - Sept. 2006
APA
2
3
4
5
6
7
8
11
12
13
15
16
19
20
21
22
23
24
25
26
27
28
30
31
32
33
34
35
36
37
38
39
40
Statewide

White
-2.1
-1.1
+0.9
-1.4
-3.5
-1.5
-0.2
-4.4
-5.5
-0.6
-0.4
+0.3
-0.7
-0.2
-0.1
-0.2
+2.1
+0.3
-2.9
-0.3
-3.3
-3.0
-2.6
+0.2
-1.4
0.0
-0.2
-0.5
-0.4
+0.2
-0.9
-0.3
-0.2

Black
+2.0
+0.4
-0.2
+2.8
+1.8
+1.1
+0.2
-0.1
-0.3
+0.1
0.0
-0.1
+0.5
0.0
-0.5
+0.1
-0.8
-0.2
-0.1
0.0
0.0
-0.2
+0.3
-0.3
+0.6
+0.5
0.0
+0.1
+0.2
-0.7
-0.1
0.0
-0.2

Native Amer.
+0.2
0.0
+0.1
0.0
+0.1
0.0
-0.2
+2.0
+0.4
0.0
+0.3
-0.1
-0.1
0.0
0.0
+0.4
0.0
0.0
-0.1
0.0
+2.2
0.0
+0.6
+0.1
+0.1
-0.1
0.0
+0.1
0.0
+0.2
+0.3
0.0
0.0

Asian
-0.4
0.0
-0.2
-1.1
+1.1
+0.4
-0.6
-0.4
-0.2
0.0
-0.1
-0.2
+0.1
+0.1
-0.2
0.0
-0.9
-0.5
-0.3
-0.4
-0.1
-0.4
-2.7
-2.0
+0.2
-1.4
0.0
-0.4
-0.1
-0.4
-0.3
-0.1
-0.1

-1.7

+0.8

+0.1

0.0

12

E. Indian Hispanic
+0.1
+0.2
0.0
+0.8
-0.2
-0.1
-0.5
+0.3
+0.1
+0.8
-0.1
+0.2
0.0
+0.3
-0.1
+4.0
0.0
+5.9
0.0
+0.4
0.0
+0.1
0.0
0.0
0.0
+0.1
+0.2
+0.1
-0.2
0.0
+0.1
+0.4
-0.2
+0.1
-0.1
+0.7
0.0
+3.2
0.0
+0.7
0.0
+2.3
-0.2
+3.8
-0.9
+0.3
-0.4
+2.5
0.0
+0.7
-0.3
+1.1
-0.1
+0.3
-0.1
0.0
0.0
+0.2
-0.1
+1.1
-0.3
+1.0
-0.1
-0.4
0.0
-0.1
0.0

+1.0

 Accidents
Arguably the most effective denominator benchmark is to compare traffic stop
data with rates of involvement in roadway accidents. These accident data can be seen as
measuring both the quantity and quality of driving in a particular area. Most importantly,
traffic accident data constitute another “blind” measure since WSP Troopers do not know
the race of those citizens they will contact in a traffic accident setting prior to arriving at
the scene of the collision or other type of accident. At a later point in this report findings
are reported from an assessment of the accuracy of traffic accident data coded for race.
The accident data are assessed against evidence collected in an observational study
permitting the coding of digital facial images of drivers on two principal roadways in the
Seattle and Spokane areas. That assessment strongly suggests that racially coded traffic
accidents likely constitute a very good estimator for the racial composition of the driving
population. These findings from an observational study – in this case the coding of facial
digital images by a group of racially and ethnically diverse coders.
(Tables 6 and 7 on the following two pages)
Table 6 displays findings on the percent of drivers contacted by the WSP as a
result of their involvement in motor vehicle accidents by race and APA, and the figures
presented in Table 7 subtract the percentage involved in accidents (by race) from the
percentage contacted as a result of self-initiated activity by the WSP. The results in Table
7 reveal that there is not a single APA in which Blacks, Native-Americans, Asians/Pacific
Islanders, East Indians, or Hispanics are over-represented in contacts as compared to
accident data (in fact, there are three APAs in which Hispanics are under-represented).
Table 6: Contacts Via Collisions (%)
Data for Nov. 2005 - Sept. 2006
13

 APA
2
3
4
5
6
7
8
11
12
13
14
15
16
19
20
21
22
23
24
25
26
27
28
30
31
32
33
34
35
36
37
38
39
40
Statewide

White
76.9
83.7
88.0
70.9
65.9
74.4
85.0
61.7
42.5
72.2
86.2
96.9
85.4
94.0
91.5
87.6
85.0
87.4
88.2
75.8
76.1
78.2
72.9
83.6
84.9
90.7
81.3
86.6
92.0
90.3
85.8
89.5
94.0
95.3

Black
8.4
6.2
2.9
7.2
9.2
3.9
0.0
1.2
0.4
1.4
0.5
0.0
3.5
1.2
3.0
2.4
1.5
1.6
2.3
1.0
3.0
0.9
1.7
1.6
1.8
2.7
3.1
1.2
1.1
2.8
1.4
0.3
0.5
0.0

Native Amer.
0.2
0.3
0.4
0.1
0.3
0.2
2.9
4.8
1.4
0.3
0.5
2.2
0.5
1.0
0.0
0.0
1.5
0.2
0.0
0.0
0.4
2.4
0.2
1.7
0.9
0.0
0.4
0.0
3.9
0.7
1.9
1.3
0.0
1.1

Asian
6.9
3.9
3.8
11.9
10.1
10.2
2.9
0.7
1.4
1.8
1.8
0.0
1.5
0.9
2.0
3.3
2.3
2.8
2.3
2.1
4.1
0.5
1.9
2.8
3.5
2.2
5.8
3.1
1.5
2.5
1.9
1.8
0.9
0.0

E. Indian
0.6
0.1
0.5
2.6
2.3
3.0
0.0
0.5
0.4
0.3
0.9
0.0
1.0
0.3
0.5
0.8
0.0
0.9
2.0
1.3
1.0
0.5
0.9
2.8
1.1
0.4
2.1
0.7
0.0
0.1
0.0
0.0
0.0
0.0

Hispanic
6.5
5.4
3.8
6.7
11.3
7.4
7.9
30.7
53.6
23.7
9.6
0.9
7.5
2.4
3.0
5.6
9.8
6.3
5.1
19.6
15.1
17.5
22.4
6.9
7.6
4.0
6.7
8.5
1.5
3.2
8.2
6.8
4.6
3.7

N
2,462
1,880
1,371
3,449
3,019
2,285
140
579
280
722
218
226
199
1,396
199
1,236
133
429
391
520
677
211
468
708
543
226
2,530
885
464
1,134
366
380
216
190

79.3

4.2

0.6

5.6

1.3

8.6

30,132

14

 Table 7: Self-Initiated Contacts Minus Contacts Via Collisions (%)
Data for Nov. 2005 - Sept. 2006
APA
2
3
4
5
6
7
8
11
12
13
14
15
16
19
20
21
22
23
24
25
26
27
28
30
31
32
33
34
35
36
37
38
39
40
Statewide

White
-2.6
-1.6
-2.7
-2.6
+0.2
+0.2
+5.3
-1.6
+7.0
+1.5
-1.7
+0.1
+2.7
+0.1
+2.1
-1.4
+3.9
+0.4
-2.8
+4.7
+9.5
+1.8
+4.6
-2.6
-1.2
-2.6
-2.7
+3.4
+0.6
-4.1
+2.6
0.0
+0.1
-0.7

Black
+3.3
+1.2
+1.6
+3.7
+3.5
+1.6
+1.7
+0.7
+0.8
+0.8
+0.5
+0.4
-1.1
+1.0
-1.2
+1.0
-0.7
+1.5
+0.4
-0.5
-0.3
-0.5
+0.4
+0.7
+0.3
+2.0
+1.2
+0.6
+0.2
+2.4
+0.7
+1.3
+0.3
+0.5

Native Amer.
+0.2
-0.1
+0.2
+0.1
0.0
+0.1
+1.9
+1.4
+0.4
-0.1
0.0
-0.7
0.0
-0.1
+0.3
+0.2
-0.4
-0.1
+0.2
+0.2
+0.1
+1.7
+0.1
-0.1
-0.5
+0.3
0.0
+0.1
-2.1
-0.3
-0.9
+0.2
+0.2
-0.2

Asian
-0.5
+0.1
+0.3
-1.7
-1.1
-1.6
-0.3
+0.7
-0.4
-0.3
-1.0
+0.3
+1.0
+0.2
+0.5
-0.1
-1.4
+0.3
+1.6
-1.0
-1.1
0.0
+0.1
+4.4
+0.9
+0.8
+1.2
-0.5
+0.6
+1.5
+0.7
-0.2
+0.7
+0.9

E. Indian
0.0
+0.3
+0.1
-0.4
+0.6
-0.5
+0.4
-0.1
-0.1
-0.1
-0.6
+0.1
-0.3
-0.2
-0.2
0.0
+0.3
-0.1
-0.8
-0.9
-0.3
-0.3
-0.3
+0.5
+0.3
-0.2
+0.1
-0.2
+0.4
+0.1
+0.3
+0.1
+0.2
+0.2

Hispanic
-0.5
+0.1
-0.6
+0.8
-2.0
+0.7
-4.3
-0.5
-7.8
-1.8
+3.0
-0.3
-1.8
-0.8
-1.7
+1.0
-2.0
-1.7
+1.4
-2.4
-7.9
-4.7
-5.2
-2.8
+0.2
-0.3
+0.5
-3.6
+0.3
+0.5
-2.9
-1.3
-1.6
-0.4

+3.8

-0.5

0.0

-2.0

-0.4

-0.9

15

 Daylight Stops
A logical argument would suggest that if racial profiling were in fact occurring, it
would be more likely to manifest itself in daylight stops than night-time stops because
WSP Troopers would be better able to form an impression of the race of individual
drivers than at times of the day when their visibility is likely to be impaired.
(Table 8 on following page)
While it is true that there may be differences in driving times and habits according
to race which traffic stop data analyzed in this manner cannot address, Table 8 presents
findings on the percentage of stops made in daylight hours2 by race for each APA. These
analyses reveal that, while there is considerable variation in the overall proportion of
daylight stops across autonomous patrol areas, (adhering to the five percentage point
criterion) a higher proportion of Blacks than Whites are stopped in daylight hours in four
APAs (APA 12 - Sunnyside; APA 23 - Kelso; APA 37 - Hoquiam; APA 39 - Raymond). A
higher proportion of Native-Americans than Whites are stopped in daylight hours in five
APAs (APA 4 - Thurston County; APA 5 - Seattle North; APA 13 - Kennewick; APA 15 Colville; and APA 21 - Vancouver). A higher proportion of Asians/Pacific Islanders than
Whites are stopped in daylight hours in three APAs (APA 27 - Okanogan; APA 30,
Bellingham; and APA 39 - Raymond). A higher percentage of East Indians than Whites
are stopped in four APAs (APA 11 - Yakima; APA 28 - Ephrata; APA 30 –
2 These data were coded such that 7 p.m. to 7 a.m. constituted non-daylight stops. While
we realize full well that there are substantial monthly/seasonal differences in the number
of daylight hours in any given day, there were no substantial differences in the number of
stops over the various months included in the data set. The coding of this variable thus
assumes that monthly/seasonal differences in the number of daylight hours will
essentially cancel each other out.
16

 Table 8: Daylight Stops (%)
Data for Nov. 2005 - Sept. 2006
APA
Overall
2
3
4
5
6
7
8
11
12
13
14
15
16
19
20
21
22
23
24
25
26
27
28
30
31
32
33
34
35
36
37
38
39
40

White

Black

Native Amer.

Asian

E. Indian

Hispanic

59.3
59.1
56.3
56.4
57.9
61.5
78.6
67.7
77.3
64.2
72.9
64.4
81.2
55.7
71.4
58.1
80.1
57.3
52.9
68.9
71.5
77.0
75.2
53.9
59.4
52.2
61.5
51.3
70.8
55.4
67.0
46.7
74.1
81.6

51.7
48.6
42.7
47.0
48.3
49.7
63.3
56.3
84.4
52.7
60.7
51.5
83.0
37.1
57.8
47.4
74.5
72.1
40.6
70.8
67.8
79.4
70.4
43.3
53.2
29.2
46.4
42.4
62.7
38.6
73.4
41.1
85.1
84.4

52.5
40.0
64.9
65.4
59.3
44.4
63.9
57.2
62.9
80.0
70.0
69.7
78.0
53.8
72.0
72.2
68.4
60.0
32.0
64.4
62.0
67.0
77.4
43.0
43.5
27.3
53.2
41.7
75.5
42.6
54.9
50.7
70.6
77.8

51.6
48.6
49.3
51.2
50.0
52.8
77.2
61.4
77.4
49.3
67.1
51.9
75.3
43.4
53.8
47.0
86.2
56.7
40.9
72.4
71.9
86.8
78.9
63.9
57.7
32.0
57.4
45.7
66.3
45.6
62.1
40.8
82.5
89.7

56.1
53.6
53.8
44.5
47.3
49.5
86.7
74.0
80.0
56.8
74.3
33.3
84.7
44.1
65.2
43.7
70.0
61.5
43.5
68.7
70.8
58.8
84.5
62.1
51.4
47.6
57.2
53.8
75.4
55.4
61.0
33.3
81.0
75.0

52.1
51.1
48.7
47.6
47.1
52.3
68.3
54.5
67.7
51.0
58.0
61.7
75.3
44.2
66.3
46.7
75.9
49.5
40.1
49.8
65.3
67.1
62.8
52.9
34.0
41.0
47.9
38.0
63.0
49.7
52.7
41.8
55.1
72.5

17

57.4
57.4
55.2
53.9
54.8
59.0
77.8
62.8
72.8
60.8
70.8
64.4
80.8
54.9
70.6
56.7
79.6
57.4
51.1
65.7
70.9
75.3
73.1
54.4
57.0
50.0
59.4
50.3
70.6
53.8
66.1
46.3
73.7
81.4

 Bellingham; APA 39 - Raymond). While these disparities should be noted, overall, this
comparison of the proportion of minority compared to white drivers who are contacted by
the WSP in daylight hours indicates that for the most part, minorities tend to be underrepresented in daylight stops.
Conclusions Regarding Stops Based on Multiple Denominators
Considered in their totality, these four distinct benchmark data comparisons
indicate rather clearly that WSP Troopers are not engaged in systemic racial profiling at
the level of which drivers they contact. This statement applies both with respect to
statewide figures, and with respect to the situation in the 34 Autonomous Patrol Areas
(APAs) distributed across the state.

The Analysis of Citations for Evidence of Biased Policing
In addition to manifesting itself in decisions to stop motorists, biased policing can
occur at the level of which drivers who are stopped by police officers are issued citations.
In this section of the 2007 traffic stop data monitoring report we present both bivariate
and multivariate statistical analyses of the citation decisions made by WSP Troopers in
the period November 2005 through September of 2006.
Table 9 presents findings on the percentage of those contacted in each APA who
were issued citations, broken down by race. The findings set forth in this table indicate
that Black drivers were somewhat more likely to be issued citations than White motorists
in 23 of 34 APAs, Native-Americans were more likely to be issued citations than Whites
in 29 APAs, Asians in 20 APAs, East Indians in 20 APAs, and Hispanics in 29 APAs.
(Table 9 on the following page)

18

 Table 9: Percent Issued Citations by Race and APA
Data for Nov. 2005 - Sept. 2006
APA
52.9
43.7
49.2
47.4
46.0
51.1
40.8
39.6
38.5
51.5
38.4
40.1
61.0
41.1
42.5
48.7
32.5
47.7
50.9
30.3
42.8
23.3
50.6
44.3
54.2
41.1
55.9
37.8
51.4
41.7
51.2
47.7
62.8
41.9
50.1

White
52.2
43.6
49.0
49.9
49.5
54.7
58.3
42.3
43.8
52.2
43.8
42.4
65.3
38.8
40.1
51.4
41.8
60.4
56.2
39.3
49.2
38.2
58.4
42.6
53.2
29.0
53.4
50.0
56.8
41.1
61.8
47.9
74.6
56.3
49.5

Black

Native Amer. Asian
E. Indian
Hisp.
65.6
51.5
43.9
59.7
56.7
44.1
53.6
54.7
72.3
48.2
49.0
60.3
57.7
44.9
42.0
50.8
66.7
45.7
43.5
59.3
55.6
50.2
47.3
53.6
55.6
53.3
53.3
46.8
36.8
41.8
42.0
48.5
28.9
34.0
20.0
52.9
70.0
41.1
38.6
57.2
50.0
38.8
57.1
54.5
45.1
55.6
33.3
59.6
71.2
72.7
70.8
68.6
54.3
39.8
32.4
47.8
48.0
46.7
60.9
50.0
36.1
44.8
41.4
55.0
48.7
44.6
20.0
42.3
60.0
58.2
57.1
54.8
60.0
61.6
54.3
60.9
40.0
33.7
32.8
32.7
50.0
50.3
55.4
47.9
35.5
31.6
52.8
39.9
47.2
62.1
60.8
48.3
59.1
62.4
60.5
54.8
67.4
73.0
69.2
56.4
54.5
28.5
42.9
38.5
57.4
61.1
58.7
61.7
43.2
47.3
48.3
48.0
57.3
64.7
66.2
58.9
59.6
36.8
28.6
46.0
51.3
56.1
58.5
48.1
58.2
46.1
50.0
55.1
64.7
75.9
85.7
61.0
33.3
48.3
33.3
56.9
53.2

47.3

50.0

19

Overall
53.2
44.5
49.9
47.6
47.7
51.3
42.0
42.3
45.0
52.6
40.7
40.4
62.0
41.3
42.7
49.0
33.7
49.0
52.2
30.8
43.7
26.1
50.7
46.9
55.5
40.1
56.6
52.1
41.6
51.4
48.2
63.1
42.5

 While these data could be interpreted as an indication that WSP troopers are more
likely to issue citations to members of minority groups than to Whites, there are a number
of important differences across the racial groups with respect to both the number of
violations observed as a result of a traffic stop and the seriousness of those violations
which influence the decision of WSP Troopers to issue citations.
Table 10 presents findings on the average number of violations of those contacted
by the WSP by race for the state’s 34 APAs. At the statewide level Whites have an
average of 1.54 violations per contact; Blacks 1.74; Native-Americans 1.87; Asians 1.48;
East Indians 1.39, and Hispanics 1.76. African Americans have a higher average number
of violations than Non-Hispanic Whites in 32 of 34 APAs, Native-Americans have a
higher average number of violations than Non-Hispanic Whites in 31 APAs, Asians and
East Indians have a higher average number of violations than Non-Hispanic Whites in
only five APAs, while Hispanics have a higher average number of violations than NonHispanic Whites in all 34 APAs.
(Table 10 and Table 11 on the following two pages)
Table 11 presents findings on the average violation seriousness score1 by race for
each of the 34 APAs. Statewide, those drivers identified as East Indian by WSP Troopers
have the lowest average seriousness scores at .08, followed by Asians at .11, NonHispanic Whites at .15, African Americans at .24, and Hispanics drivers at .25. Native1 This variable is coded “one” for serious offenses and coded “zero” for other offenses,
and then summed across the eight violation fields (with possible scores ranging from zero
to eight). Serious violations included the following offenses: felony drugs; misdemeanor
drugs; DUI drugs with test; DUI drugs, no test; DUI underage, with test; DUI underage,
no test; DUI with test; DUI without test; felony flight, elude; felony warrant; hit and run;
insurance - none; license suspension/revocation; misdemeanor warrant; negligent driving,
1st degree; negligent driving, 2nd degree; reckless driving; vehicular homicide; and
vehicular assault.
20

 Table 10: Average Number of Violations by Race and APA
Data for Nov. 2005 - Sept. 2006
APA
2
3
4
5
6
7
8
11
12
13
14
15
16
19
20
21
22
23
24
25
26
27
28
30
31
32
33
34
35
36
37
38
39
40

White
1.63
1.78
1.80
1.65
1.62
1.63
1.74
1.39
1.60
1.39
1.54
1.66
1.35
1.53
1.28
1.56
1.36
1.62
1.37
1.76
1.58
1.39
1.45
1.59
1.54
1.46
1.54
1.39
1.50
1.60
1.29
1.48
1.38
1.43

Statewide 1.54

Black
1.87
1.97
1.83
1.90
1.83
1.87
1.85
1.48
1.75
1.45
1.68
1.97
1.43
1.72
1.36
1.82
1.42
1.66
1.36
1.94
1.60
1.44
1.56
1.78
1.49
1.52
1.64
1.44
1.70
1.70
1.32
1.51
1.39
1.44

Native Amer.
2.10
2.47
2.30
1.81
2.07
1.96
2.19
1.76
1.92
1.33
1.64
1.91
1.41
1.81
1.56
1.36
1.74
2.80
2.00
1.87
1.99
1.90
1.70
2.10
2.15
1.55
1.92
1.42
1.96
2.04
1.55
2.09
1.47
1.22

Asian
1.57
1.79
1.70
1.59
1.57
1.61
1.61
1.32
1.40
1.50
1.36
1.59
1.23
1.58
1.35
1.50
1.34
1.46
1.25
1.75
1.58
1.39
1.34
1.26
1.26
1.38
1.38
1.29
1.41
1.47
1.30
1.37
1.26
1.34

1.74

1.87

1.48

21

E. Indian
1.52
1.77
1.49
1.51
1.46
1.55
1.93
1.42
1.80
1.32
1.31
1.78
1.17
1.44
1.17
1.26
1.20
1.32
1.14
1.51
1.58
1.29
1.26
1.31
1.16
1.48
1.30
1.20
1.35
1.59
1.12
1.00
1.24
1.00

Hispanic
1.87
1.98
2.05
1.90
1.88
1.87
1.82
1.66
1.96
1.64
1.76
1.94
1.52
1.68
1.40
1.63
1.37
1.81
1.55
2.03
1.75
1.72
1.74
1.76
1.82
1.56
1.73
1.51
1.72
1.69
1.36
1.59
1.63
1.53

1.39

1.76

 Table 11: Average Seriousness of Violations by Race and APA
Data for Nov. 2005 - Sept. 2006
APA
2
3
4
5
6
7
8
11
12
13
14
15
16
19
20
21
22
23
24
25
26
27
28
30
31
32
33
34
35
36
37
38
39
40
Statewide

White
.23
.26
.22
.20
.19
.13
.17
.10
.09
.11
.14
.13
.09
.15
.07
.16
.09
.19
.12
.16
.11
.09
.09
.16
.19
.16
.16
.12
.13
.15
.09
.21
.13
.13
.15

Black
.32
.34
.21
.30
.29
.23
.32
.14
.13
.14
.22
.09
.13
.23
.12
.26
.11
.19
.14
.28
.16
.18
.13
.21
.15
.17
.20
.12
.25
.18
.10
.22
.06
.28
.24

Native Amer.
.57
.67
.55
.23
.63
.33
.44
.31
.22
.10
.30
.30
.14
.34
.24
.14
.29
.70
.24
.36
.20
.35
.17
.52
.52
.50
.34
.33
.26
.43
.21
.45
.12
.11
.34

Asian
.18
.20
.14
.15
.15
.10
.22
.04
.09
.09
.11
.19
.06
.12
.06
.11
.05
.12
.05
.09
.08
.13
.03
.04
.05
.09
.08
.07
.08
.11
.08
.16
.07
.05
.11

22

E. Indian
.13
.39
.08
.13
.15
.07
.27
.06
.13
.05
.20
.11
.04
.12
.04
.05
.10
.05
.07
.04
.08
.00
.07
.04
.04
.19
.06
.04
.08
.16
.00
.00
.00
.08

Hispanic
.35
.39
.32
.34
.36
.25
.25
.25
.25
.24
.30
.23
.15
.23
.13
.19
.14
.25
.21
.25
.21
.26
.22
.27
.37
.21
.28
.16
.17
.21
.12
.22
.32
.21

.08

.25

 American drivers contacted by the WSP had the highest average seriousness scores,
registering at .34. At the individual APA level, Blacks and Native-Americans had
average seriousness scores that were lower than those of Whites in only four of thirtyfour APAs; Asians had average seriousness scores that were higher than Non-Hispanic
Whites in only three APAs, while East Indians had average seriousness scores that were
higher than Non-Hispanic Whites in seven APAs. Hispanic drivers had higher average
seriousness scores than Non-Hispanic Whites in all 34 APAs.

These cross-race

differences in the number of violations observed during a traffic stop, and in the
seriousness of violations noted are taken into account in the multivariate analyses of
citation decisions of WSP Troopers presented below.
Multivariate Analysis of Citations
Our multivariate analyses of citations focus on two dependent variables: (1)
whether an individual contacted by the WSP was issued a citation as a result of the traffic
stop, and (2) in situations of multiple violations, the number of citations issued. Taking
into account the points made above regarding differences in the average number and
seriousness of violations across racial groups, we conducted separate analyses for each of
the 34 autonomous patrol areas with the predictor/independent variables in the first model
consisting of the driver’s gender (males coded one, females coded zero); age (in years, a
continuous variable); and race (dummy variables for African American, NativeAmerican, Asian, East Indian, and Hispanic, with Non-Hispanic Whites treated as the
reference category). We also included measures of the number of violations of the
individual contacted and the combined seriousness of those violations; a variable
indicating whether the stop occurred during daylight hours; and a variable indicating

23

 whether the stop occurred on an interstate highway (interstate highway coded one; all
other locations coded zero). We also included separate “dummy variables” (binary
variables indicating the presence or absence of a trait or characteristic) for individuals
contacted who had “out-of-state” license plates (most typically from British Columbia,
California, and Oregon).
The second set of multivariate statistical models developed for the analysis
include all the variables identified above, as well as interaction terms for race multiplied
by the number and combined seriousness of the violations in order to control for the
possible effects on being issued a citation of differences in rates of noncompliance with
traffic laws across racial groups.
While our focus in this report is on the impact of race on being issued a citation,
the full logistic regression models run on the WSP traffic stop data for the period
November 2004 through September 2006 indicated with respect to gender that males
were significantly more likely to be issued a citation in 21 of 34 APAs (see Table 12); age
had a statistically significant impact on receiving a citation in 32 of the 34 APAs (with
younger drivers being more likely to be issued a citation); stops occurring on interstate
highways were more likely to result in citations being issued in 15 APAs; and daylight
stops were more likely to result in citations being issued in 30 APAs. The findings set
forth in Table 13 indicate that the number of violations had a statistically significant
impact on receiving a citation in 15 APAs (those with a greater number of violations were
more likely to be issued citations), and the combined seriousness score had a statistically
significant effect on receiving a citation in 33 APAs (those with higher
(Table12 and Table 13 on the following two pages)

24

 Table 12: Odds Ratios - Citation Dependent Variable (with Interaction Terms)
Data for Nov. 2005 - Sept. 2006
APA
2
3
4
5
6
7
8
11
12
13
14
15
16
19
20
21
22
23
24
25
26
27
28
30
31
32
33
34
35
36
37
38
39
40

Male
1.11**
1.18**
1.12**
1.33**
1.33**
1.12**
1.69**
1.26**
.98
.98
1.08
1.16*
.86**
1.19**
1.05
1.15**
1.16
1.01
.91
1.25**
.90*
1.37**
1.08
1.43**
1.17**
1.11
1.10**
1.14**
1.07
1.11**
1.03
1.05
1.09
1.21*

Age
.99**
.99**
.99**
.98**
.98**
.98**
.97**
.98**
.98**
.99**
.99**
.99**
.98**
.99**
.99**
.99**
.98**
.98**
.98**
.99**
.98**
.98**
.98**
.98**
.99**
1.00
.98**
.99**
.99**
.99**
.98**
.99**
1.00
.98**

Interstate
.95
.92
1.44**
1.08
1.14**
1.05
4.13**
1.11
1.11
1.26**
1.53
N.A.
1.84**
1.62**
N.A.
1.28**
1.84
1.38**
1.45**
.32
1.89**
N.A.
1.88**
1.83**
1.66**
.38
2.00**
1.00
N.A.
.23**
N.A.
4.03**
N.A.
.92

* p < .01
** p < .001

25

Daylight Stop
1.37**
1.56**
2.12**
1.92**
1.46**
1.91**
1.33*
2.22**
1.71
2.47**
1.97**
2.40**
1.58**
2.39**
3.15**
2.08**
2.07**
2.75**
1.38**
2.59**
1.97**
1.22*
2.29**
2.80**
2.77**
4.64**
2.41**
5.40**
1.81**
2.76**
3.46**
1.48**
4.01**
1.50**

 Table 13: Odds Ratios - Citation Dependent Variable (with Interaction Terms)
Data for Nov. 2005 - Sept. 2006
APA
2
3
5
6
7
8
11
12
13
14
15
16
19
20
21
22
23
24
25
26
27
28
30
31
32
33
34
35
36
37
38
39
40

# Violations Seriousness BC plate
CA plate
OR plate
1.01
4.03**
.86
1.04
1.15
1.16**
4.37**
.42
1.04
.87
1.14**
4.05**
2.14**
1.79**
1.65**
.95
4.81**
1.58*
.91
1.15
1.16**
3.12**
1.63
.95
1.07
.98
4.01**
1.09
1.16
.90
1.14
2.06**
.19
1.00
.69
1.16**
4.16**
.95
1.66**
1.25
.91
5.59**
.49
1.10
.67
.94
5.11**
1.38
1.54**
1.26**
1.22**
1.72**
1.27
1.23
1.10
1.26**
3.62**
2.81**
2.75**
1.41
.91
4.32**
1.47
1.05
1.19
.96
5.70**
1.40
1.57*
.97
.75
3.26**
1.05
2.32**
1.58*
1.03
4.33**
.88
1.28
1.21**
1.18**
2.07**
.98
1.70**
1.34**
1.01
3.75**
1.60*
2.25**
1.31**
1.23**
1.57**
1.61*
1.39**
1.36**
1.19**
2.22**
1.31
1.83**
1.32
.73
3.75**
1.49*
1.30
.76
1.14
3.92**
1.06
2.39**
1.18
.76
4.60**
1.10
1.56*
1.07
.94
5.11**
1.60**
.64
1.32
1.03
4.66**
1.96**
1.82*
.87
1.07
5.78**
1.59
1.60*
2.14**
2.41**
.94
3.79**
2.01**
1.32*
1.16**
6.57**
1.23
.69
1.10
3.88**
3.22**
1.25
1.46**
1.65**
1.15**
3.46**
.59
.92
.86
1.00
5.58**
1.75
1.27
1.25
1.48**
2.69**
1.51
1.23
1.76
1.07**
2.35**
2.61**
1.60*
.91
1.14
1.58**
1.99
2.01
1.59**

* p < .01
** p < .001

26

 seriousness scores were more likely to receive citations). Drivers with BC plates were
more likely to be issued citations in six APAs; those with California plates were more
likely to be issued citations in 12 APAs, while those with Oregon plates were more likely
to be issued citations in nine APAs.
Table 14 presents summary odds ratios for the effects of race on citation (these
models included interaction terms and all other independent variables). This table reveals
that African Americans were not more likely to be issued citations in a single APA, and
were significantly less likely (p < .001) to be issued a citation in four APAs (Tacoma
Freeway, Seattle North, Everett Central, and Everett East). Native-Americans were not
more likely to be cited in a single APA, and were significantly less likely to receive a
citation in Yakima. Hispanics were not more likely to be issued a citation in a single
APA, and were significantly less likely to be issued citations in the Seattle North, Seattle
East, Wenatchee, and Hoqiuam APAs. While the results for East Indians should be
treated with caution due to the large number of APAs in which there were too low a
number of contacts with members of this group to allow for reliable statistical analyses,
there was only one APA (Bellingham) in which East Indians were more likely to be
issued citations than the comparison group of Non-Hispanic Whites. However, these
analyses did indicate that Asian drivers were significantly more likely to be issued
citations in five of the state’s APAs, those being Kelso, Ellensburg, Bellingham, Mount
Vernon, and Everett Central.
(Table 14 on the next page)
The disproportionate citation rate for Asian drivers in these five APAs appears to
be related to higher rates of citation for members of this group as a result of contacts for

27

 Table 14: Odds Ratios - Citation Dependent Variable (with Interaction Terms)
Data for Nov. 2005 - Sept. 2006
APA
2
3
4
5
6
7
8
11
12
13
14
15
16
19
20
21
22
23
24
25
26
27
28
30
31
32
33
34
35
36
37
38
39
40

Black
.70**
.79
1.01
.64**
.85
.80
.62
1.08
.59
.83
1.76
1.47
.79
.55*
.55
.69
.72
.90
1.06
1.44
1.05
2.05
.94
.56
1.03
.67
.64**
.40**
.82
.77
1.78
1.19
1.39
1.00

Native
.60
.41
1.72
.77
1.81
1.96
.15
.38**
.17*
3.50
1.51
.86
2.63
.62
.19
.12
.85
.17
.30
.62
1.02
1.05
.10*
1.46
.51
1.75
.75
N.A.
1.11
1.26
.41
.37
1.28
1.63

Asian
1.24
.98
1.34
.95
1.19
1.10
1.90
1.83
.88
.74
.79
3.04
2.22*
.90
1.25
1.17
1.02
2.53**
1.22
1.76
1.69**
1.46
1.69
3.11**
4.00**
1.06
1.71**
1.49
1.99*
1.06
1.51
.15
1.09
1.34

* p < .01
** p < .001

28

Hispanic
.57
.91
1.27
.58**
1.08
.64**
.54
.88
1.01
.85
1.02
2.86
1.41
.76
.65
.87
.83
.87
1.04
.62**
.98
1.29
.55
.96
.70
.85
.87
.71
1.50
.89
.50**
.35*
.77
.74

East Indian
.87
1.09
.43
.86
1.03
1.18
.29
1.15
.65
.30
1.37
1.52
.43
1.25
4.58
.50
1.20
1.05
.44
1.86
1.72
3.27
.89
2.15**
1.85
.71
.87
.80
2.04
.56
3.44
1.84
N.A.
1.00

 speeding violations. In APA 23 (Kelso), 78.2% of Asian drivers who were contacted as a
result of speeding (via radar patrols) were issued citations, compared with 67.4% of nonAsian drivers. In APA 26 (Ellensburg) 61.1% of Asian drivers contacted as a result of
speeding (radar) were issued citations, compared to 51.7% of non-Asian drivers. In APA
30 (Bellingham) 72.9% of Asians contacted as a result of speeding (radar) were issued
citations, compared with 53.3% of non-Asian drivers. In APA 31 (Mount Vernon) 86.7%
of Asian drivers contacted as a result of speeding (radar) were issued citations, compared
to 60.4% of non-Asian drivers. In APA 33 (Everett Central) 84.5% of Asian drivers
contacted as a result of speeding (radar) were issued citations, compared to 76.8% of nonAsian drivers.
Critics of our multivariate analytical approach might contend that our finding of
attenuated racial/ethnic bias in the issuing of citations when the number and seriousness
of violations across racial/ethnic groups is considered is an artifact which itself is the
result of racial bias on the part of members of the Washington State Patrol. If officers
record a greater number and severity of violations for members of minority groups, this
could be the product of bias rather than the actual driving behavior of those contacted. In
order to address this potential criticism, we conducted an additional set of analyses on the
probability of receiving a citation for each of the 34 APAs for drivers who had only one
recorded violation. These analyses included all variables (with the obvious exception of
the number of violations and the interaction terms) included in previous models.

(Table 15 on the following page)

29

 Table 15: Odds Ratios - Citation Dependent Variable (Single Violation Recorded)
Data for Nov. 2005 - Sept. 2006
APA
2
3
4
5
6
7
8
11
12
13
14
15
16
19
20
21
22
23
24
25
26
27
28
30
31
32
33
34
35
36
37
38
39
40

Black
.79**
.82
.94
.82
.95
.97
.87
1.10
.50
1.09
1.61
1.26
1.06
.63**
.74
.89
1.38
.98
1.17
1.59
1.15
1.91
1.23
.66
.91
.71
.81*
.63*
.93
.93
1.48
1.08
1.44
1.23

Native
.89
.35
2.36
.78
2.45
2.18
.70
.45**
.32
3.04
1.58
.74
2.01
.98
.49
.29*
1.22
.55
.55
1.03
1.15
1.06
.33*
1.25
1.09
1.50
.96
N.A.
1.15
1.64
.80
.68
1.60
1.07

Asian
1.13
1.06
1.09
.97
1.10
1.06
1.78
1.29
.71
.69
1.00
2.57
1.69
.94
1.45
.97
1.38
1.53*
1.30
1.33
1.34*
1.74
1.59**
1.84**
2.20**
.93
1.26**
1.09
1.88**
.99
1.33
.78
1.57
1.31

* p < .01
** p < .001

30

Hispanic
1.05
1.12
1.41*
.74*
1.26*
.84
.79
1.06
1.41**
1.02
1.37**
3.15*
1.38
.98
1.09
1.07
1.10
.88
1.22
.83
1.06
1.40**
.78**
1.12
.91
.91
1.11
.93
1.64
.93
.72**
.82
.97
1.08

East Indian
.78
1.06
.65
.87
.96
.98
.70
1.44
.24
.84
3.08
.88
2.25
1.00
2.90
.88
.69
1.30
1.04
1.21
1.99
3.41
1.09
1.65**
1.53
.79
1.01
1.21
1.56
.63
1.99
1.94
N.A.
.70

 Table 15 reveals that African American drivers who had only one recorded
violation were not significantly more likely to be issued a citation in a single APA, and
were significantly less likely to be issued a citation in the Tacoma Freeway and Spokane
APAs. Native-Americans with a single recorded violation were not significantly more
likely to be issued a citation in any APA, and were significantly less likely to be issued a
citation in the Yakima APA. Hispanics with a single violation were significantly more
likely to be cited in the Sunnyside, Walla Walla, and Okanogan APAs, and were
significantly less likely to be cited in Bellingham and Hoqiuam. East Indian drivers with
a single recorded violation were significantly more likely to be cited in the Bellingham
APA, while Asian drivers with a single violation were significantly more likely to be
cited in the Ephrata, Bellingham, Mount Vernon, Oak Harbor and Port Angeles APAs.
While these multivariate statistical analyses reveal somewhat more evidence of potential
bias in additional APAs across the Evergreen state, they do not indicate the existence of
systematic bias in citing minorities who have a single violation recorded by the
Washington State Patrol.
In order to examine an additional potential manifestation of bias (namely, the
“piling on” phenomenon whereby the police are engaged in issuing a greater number of
citations to members of minority groups than they do to Non-Hispanic Whites for the
same type of offenses) a final set of analyses of citations issued in the Nov. 2005 to
September 2006 period was done. The research team selected those cases in which more
than a single violation was recorded, and treated the number of citations issued as the
dependent variable, using ordinary least squares regression and statistically controlling

31

 for the other independent variables. Table 16 presents beta coefficients (a measure of
standardized effects under conditions of controlling for all other predictor variables) for
the race variables. This analysis indicates that there were no APAs where African
American drivers were issued a greater number of citations, and two (Bellingham and
Everett East) in which they were issued significantly fewer citations. There were no
APAs in which Native-Americans were issued a greater number of citations, and one
(Wenatchee) in which they were issued significantly fewer citations. Hispanics were
issued a greater number of citations in Goldendale, and significantly fewer citations in
Seattle East, Walla Walla, Wenatchee, and Ellensburg. East Indians were not issued more
citations in a single APA, while Asians with more than one recorded violation were issued
a greater number of citations in seven APAs (Thurston County, Seattle South, Kelso,
Chehalis, Mount Vernon, and Everett Central).
Considering the analyses of citations overall, while there should be some concern
regarding the higher rate of citation for Asian drivers in certain APAs, it is important to
note that when racial differences in compliance with traffic and safety laws are
statistically controlled for, there is no evidence of systematic racial bias on the part of the
Washington State Patrol at the level of which drivers are issued citations.

(Table 16 on the following page)

32

 Table 16: Beta Coefficients, with Number of Citations as Dependent Variable
(Interaction Terms Included )
Data for Nov. 2005 - Sept. 30, 2006
APA
2
3
4
5
6
7
8
11
12
13
14
15
16
19
20
21
22
23
24
25
26
27
28
30
31
32
33
34
35
36
37
38
39
40

Black
- .05
.00
.03
- .03
.00
- .01
- .11*
.06
.05
.01
.03
.02
- .03
- .03
.07
- .03
- .05
.08
.04
- .01
.02
.00
- .02
- .10**
- .03
.01
- .01
- .10**
.02
- .01
.01
.08
.05
- .01

Native
.03
- .01
.04
.02
.00
.01
- .10
- .03
- .03
- .07*
.04
.03
.00
- .01
- .10
.01
.06
.00
- .12
- .07**
- .02
.00
- .02
.04
- .05
.06
- .01
.04
- .03
.01
.02
- .03
- .04
.03

Asian
.03
- .01
.10**
.01
.09**
.05
- .05
.10
.05
.07
.02
.03
.07
.01
- .05
.06
.13
.10**
.19**
.05
.10**
.03
- .02
.11*
.17**
.03
.11**
- .03
.03
.01
.05
.12
.10
.05

* p < .01
** p < .001

33

Hispanic
- .01
- .01
- .02
- .02
-.03
- .10**
- .17*
- .01
.05
.02
- .18**
- .01
.05
.00
- .07
.05
.21**
.02
- .04
- .14**
- .09**
- .07
- .06
- .03
- .02
- .02
.02
.01
- .02
.00
.07
.06
- .06
- .03

East Indian N
- .04
6,489
.08
7,493
.04
8,022
- .01
6,579
- .01
8,712
.04
9,364
- .07
1,853
- .08
3,970
- .11*
2,891
- .06
5,896
- .04
4,709
.04
3,721
.04
3,018
.01
8,562
- .04
1,667
.06
7,919
.02
1,934
.07
4,499
.02
3,140
.07
9,601
.01
6,353
.01
2,490
.07
5,378
- .03
4,069
- .01
3,707
- .03
2,757
.11
11,605
.01
4,933
.03
6,092
- .01
10,603
- .01
3,193
N.A.
1,457
N.A.
2,399
- .03
1,849

 Comprehensive Multivariate Analysis of Citations Issued:
Drilling Down to the Contextual Detail of Citation Issuance
We conducted multivariate logistic regression analyses of the decision to issue a
citation.

In addition to the independent variables of race/ethnicity (Black, Native-

American, Asian/Pacific Islander, Hispanic, and East Indian, treating Whites as the
reference category) we included controls for gender (males=1; females =0); age (a
continuous variable) whether the traffic stop occurred on an Interstate highway
(Interstate=1; other=0); whether the stop occurred during daylight hours (daylight=1;
night=0); the number of violations identified as a result of the traffic stop (a continuous
variable); the seriousness of the violation(s); four dummy variables indicating whether
the individual contacted had an out of state license plate (British Columbia, California,
Oregon, and Idaho) and multiplicative interaction terms for each of the five minority
groups controlling for differences in the number and seriousness of violations across the
groups.
Table 17 features odds ratios for the citation decision for each of the five minority
groups for the full logistic regression models. Using a .001 probability criterion, this
table reveals that African Americans were not significantly more likely to be issued
citations in any autonomous patrol areas, to the contrary, as a group they were
significantly less likely to be issued citations than Non-Hispanic Whites in three APAs.
Similarly, Native-American drivers were not significantly more likely than Non-Hispanic
Whites to be issued citations in any of the state’s 34 APAs, and as a group they were
significantly less likely to be issued citations in one APA. Asian drivers in the state were
(Table 17 on the following page)

34

 Table 17: Odds Ratios -Citation Dependent Variable
(Interaction Terms Included)
Data for Nov. 2005 - Sept. 2006
APA
2
3
4
5
6
7
8
11
12
13
14
15
16
19
20
21
22
23
24
25
26
27
28
30
31
32
33
34
35
36
37
38
39
40

Black
.70**
.79
1.01
.64**
.85
.80
.62
1.08
.59
.83
1.76
1.47
.79
.55
.55
.69
.72
.90
1.06
1.44
1.05
2.05
.94
.56
1.03
.67
.64**
.40**
.82
.77
1.78
1.19
1.39
1.00

Native
.60
.41
1.72
.77
1.81
1.96
.15
.38**
.17
3.50
1.51
.86
2.63
.62
.19
.12
.85
.17
.30
.62
1.02
1.05
.10
1.46
.51
1.75
.75
N.A.
1.11
1.26
.41
.37
1.28
1.63

* p < .01
** p < .001

35

Asian
Hispanic
1.24
.57
.98
.91
1.34
1.27
.95
.58**
1.19
1.08
1.10
.64**
1.90
.54
1.83
.88
.88
1.01
.74
.85
.79
1.02
3.04
2.86
2.22
1.41
.90
.76
1.25
.65
1.17
.87
1.02
.83
2.53**
.87
1.22
1.04
1.76
.62**
1.69**
.98
1.46
1.29
1.69
.55
3.11**
.96
4.00**
.70
1.06
.85
1.71**
.87
1.49
.71
1.99
1.50
1.06
.89
1.51
.50**
.15
.35
1.09
.77
1.34
.74

East Indian
.87
1.09
.43
.86
1.03
1.18
.29
1.15
.65
.30
1.37
1.52
.43
1.25
4.58
.50
1.20
1.05
.44
1.86
1.72
3.27
.89
2.15**
1.85
.71
.87
.80
2.04
.56
3.44
1.84
N.A.
1.00

 significantly more likely to be issued citations in 5 of the state’s 34 APAs. Hispanics
were not significantly more likely than Non-Hispanic Whites to be issued citations in any
APA, and were significantly less likely to be issued citations in four APAs. East Indians
were significantly more likely to be issued citations in only one of the APAs.
Considered collectively, these numerous analyses conducted from a variety of
analytical perspectives indicate rather convincingly that as an agency the WSP and its
Troopers are not engaged in biased policing at the level either of traffic stops or at the
level of the issuance of citations.

The Analysis of Searches Conducted as a Result of a Traffic Stop for
Biased Policing
In this section, we replicate most our earlier statistical analyses on the relationship
between search and seizure and race using Washington State Patrol self reported data
from November 2005 through September 2006. As we have done in past reports, we
consolidate the various search categories into three distinct categories for the purposes of
analysis: No Search, Low Discretion searches and High Discretion searches.

Low

Discretion searches include searches incident to arrest, impound or inventory searches,
and warrant searches. Each of these searches is triggered by a preceding event, and thus
the decision to search involves relatively less discretion than High Discretion searches.
High Discretion searches include consent searches, “Terry” or pat-down searches, and k-9
searches. Search incident to arrest is the most common type of search conducted with
well over half of all searches fitting into this category. Terry or pat-down searches are the
most common high discretion search.

36

 Table 18 shows the frequencies associated with each of these three search
categories. As we have found in previous reports to the Washington State Patrol, searches
resulting from traffic stops continue to be highly infrequent events. Only 3.3 percent of
motorists contacted by WSP Troopers were searched during the 11-month time period
analyzed here, with 2.9 percent being subjected to a low discretion search, and only .4
tenths of a percent being subjected to a high discretion search.
Table 19 shows the results of a cross-tabulation of the search categories by race
and ethnicity. The results reported here are quite similar to our earlier analysis of the
2002 and the 2003-2004 WSP traffic stop data. As a proportion within racial groups,
Native Americans continued to be searched at higher rates than other races. Over 12
percent of Native Americans contacted by the WSP were subjected to a low discretion
search, and 1.4 percent of Native Americans pulled over for a traffic stop were subjected
to a high discretion search. Hispanics were subjected to low discretion searches 5.4
percent of the time, and were subjected to high discretion searches .9 of a percent of the
time. African Americans were subjected to low discretion searches 5.3 percent of the
time, and to high discretion searches .7 tenths of a percent of the time. Non-Hispanic
Whites were subjected to low discretion searches 2.6 percent of the time, and to high
discretion searches .3 tenths of a percent of the time. Asians/Pacific Islanders were
subjected to low discretion searches 2.3 percent of the time, and they subjected to high
discretion searches .3 tenths of a percent of the time. And lastly, East Indians were
subjected to low discretion searches .7 tenths of a percent of the time, and to high
discretion searches .1 tenth of a percent of the time.
(Table 18 and Table 19 on the following page)

37

 Table 18: Frequencies of Low and High Discretion Searches
From all observations, Nov. 2005-Sept. 2006
Frequency

Percent

Cumulative
Percent

No Search

834,778

96.7

96.7

Low Discretion Search

25,131

2.9

99.6

High Discretion Search

3,407

.4

100.0

863,316

100.0

100.0

Total

Table 19: Categories of Search by Race
From all observations, Nov. 2005-Sept. 2006*
No Search

White

691,534 (97.1%)

Low Discretion
Search
18,346 (2.6%)

Black

30,653 (94.0%)

1,719 (5.3%)

227 (0.7%)

32,599 (100%)

Native Am.

4,004 (85.9%)

590 (12.7%)

66 (1.4%)

4,660 (100%)

Asian/Pac

28,047 (97.4%)

665 (2.3%)

77 (0.3%)

28,789 (100%)

Hispanic

60,848 (93.7%)

3,506 (5.4%)

577 (0.9%)

64,931 (100%)

East Indian

15,212 (99.2%)

101 (0.7%)

20 (0.1%)

15,333 (100%)

Other

3,647 (97.1%)

98 (2.6%)

9 (0.2%)

833,945 (96.7%)

25,025 (2.9%)

3,405 (0.4%)

Total

High Discretion
Search
2,429 (0.3%)

Total
712,309 (100%)

4,225 (100%)
862,375 (100%)**

*The total number of observations is less that the total reported in Table 1 due to a relatively small number
of missing variables.
**percentages may not actually add up to 100 percent due to rounding errors.

38

 Table 20 shows the hit rates for each racial/ethnic group, for both low and high
discretion searches.

Hit rates represent the rate at which law enforcement find

contraband as a result of a search. They are by common convention calculated as a
proportion of the total “hits” of contraband within each search category and within each
racial/ethnic group category. For low discretion searches, the hit rate was .26 for NonHispanic Whites, African Americans and Native Americans, .20 for Hispanics, .17 for
Asians/Pacific Islanders, and .10 for East Indians. For high discretion searches, the hit
rate was .18 for Non-Hispanic Whites, .17 for Native Americans, .15 for African
Americans and Hispanics alike, .11 for East Indians and .10 for Asians/Pacific Islanders.
Table 21 presents the results of the multivariate analysis we did using a
multinomial logit model. The dependent variable for the model is the three category
search variable (0=No Search, 1=Low Discretion Search, 2=High Discretion Search).
The specification of the model and the independent variables used the same as those in
our previous report that analyzed the 2003-2004 data, excluding the APA or District level
variables. When controlling for these covariates, this analysis indicates that for both
categories of searches, males are more likely to be searched than females, younger
drivers are more likely to be searched than older drivers, Native Americans, Hispanics
and blacks are more likely to be searched than whites, while Asians/Pacific Islanders and
East Indians are the least likely to be searched. The nature of contact variables indicate
that the more the number of violations the more likely a search will occur, if the stop
involved a serious violation a search is more likely, searches are more likely on
interstates and non-interstates, and searches are less likely in the daylight hours than night
(Table 20 and Table 21 on the following page)

39

 Table 20: Hit Rates by Race, Across Categories of Search
From all observations, Nov. 2005-Sept. 2006
Low Discretion Searches

High Discretion Searches

White

No
Contraban
d
13,618

Black

1,278

441

.26

194

33

.15

Nat. Am.

435

155

.26

55

11

.17

Asian/Pac

550

115

.17

70

7

.10

Hispanic

2,922

584

.20

493

84

.15

E. Indian

91

10

.10

18

2

.11

77
18,971

21
6,056

.21
.27

8
2,833

1
570

.13
.17

Other
Total

Contraband

Hit
Rate

No
Contraband

Contraband

Hit
Rate

4,730

.26

1995

432

.18

Table 21: Multinomial Logit Results
From all observations, Nov. 2005-Sept. 2006
Variable

Low Discretion Search
Coefficient
Significance
(S.E.)
level

High Discretion Search
Coefficient
Significance
(S.E.)
level

Driver Characteristics:
Female
Age
Black
Hispanic
Native American
Asian/ Pacific Islander
East Indian
Other Race

-0.205 (.018)
-0.27 (.001)
0.414 (.030)
0.343 (.022)
1.417 (.055)
-0.253 (.044)
-1.735 (.112)
-0.403 (.120)

.00
.00
.00
.00
.00
.00
.00
.00

-0.747 (.048)
-0.045 (.002)
0.510 (.070)
0.588 (.048)
1.449 (.128)
-0.416 (.117)
-1.107 (.225)
-0.526 (.335)

.00
.00
.00
.00
.00
.00
.00
.12

Nature of Contact:
Number of Violations
Serious Violation(s)
Interstate
Daylight

0.687(.004)
3.416 (.021)
0.038 (.015)
-1.080 (.015)

.00
.00
.01
.00

0.355 (.012)
2.031 (.062)
0.223 (.036)
-0.508 (.035)

.00
.00

Officer Characteristics:
Female Officer
White Officer

-0.226 (.031)
0.309 (.026)

.00
.00

-0.438 (.086)
0.451 (.067)

.49
.00

.00

-4.793 (.091)

.00

Constant

-3.984 (.036)

40

.00

 time. The officer characteristics variables indicate that male officers are more likely to
conduct searches than female officers and white officers are more likely to conduct
searches than non-white or minority officers.
Because the magnitude of the coefficients produced by the multinomial logit
analysis are difficult to interpret, it is useful to employ the coefficient values to produce
predicted probabilities of searches based on these data and this model. Table 23 presents
predicated probabilities of searches for male and female drivers of different races and age
categories. The probabilities were calculated for stops involving a Non-Hispanic White
male police officer, in the daytime, on an interstate, with one (non-serious) violation. For
example, among 18-year old male drivers, Non-Hispanic Whites have a probability of .
008 of being subjected to a low discretion search and a .004 probability of being
subjected to a high discretion search; African Americans have a probability of .012 of
being subjected to a low discretion search, and a .006 probability of being subjected to a
high discretion search; Hispanics have a probability of .011 of being subjected to a low
discretion search, and a .007 probability of being subjected to a high discretion search;
Native Americans have a probability of .031 of being subjected to a low discretion
search, and they have a .016 probability of being subjected to a high discretion search;
Asians/pacific islanders have a probability of .006 of being subjected to a low discretion
search, and a .0025 probability of being subjected to a high discretion search; and East
Indians have a probability of .001 of being subjected to a low discretion search and a .001
probability of being subjected to a high discretion search. Within each racial group, the
probabilities of females being searches are lower than that of males, and 50-year olds are
less likely to be searched than 18-year olds.

(Table 22 on next page)

41

 Table 22: Predicted Probabilities of Searches*
From all observations, Nov. 2005-Sept. 2006

No Search
Age:

Low Discretionary
Search

High
Discretionary
Search
18
50

18

50

18

50

Male
White

.988

.996

.008

.003

.004

.001

Black

.982

.993

.012

.005

.006

.0015

Hispanic

.982

.994

.011

.005

.007

.0016

Native American

.953

.982

.031

.014

.016

.004

Asian

.991

.997

.006

.003

.0025

.001

East Indian

.997

.999

.001

.001

.001

.0003

Female
White

.992

.997

.006

.003

.002

.0004

Black

.987

.995

.010

.004

.003

.0007

Hispanic

.988

.995

.009

.004

.003

.0007

Native American

.966

.987

.026

.011

.008

.002

Asian

.994

.998

.005

.002

.001

.0003

East Indian

.998

.999

.001

.0004

.0006

.0001

*Predicted probabilities were calculated for stops involving a white male police officer,
in the daytime, on an interstate, with one (non-serious) violation.

42

 To summarize the findings reported here, they are remarkably consistent with the
results of our analysis of searches using the 2003-2004 data in our previous report to the
Washington State Patrol. Among other things, this consistency in findings indicates that
there has not been any noteworthy “de-policing” as a result of the WSU research team’s
ongoing monitoring of WSP traffic stop data and earlier findings of disparities among the
search rates of different races. The behavior of the WSP in conducting searches appears
to have been essentially the same in 2006 as it was in the 2003-2004 period analyzed in
previous reports to the WSP. There remains a correlation between the race of the driver
and the likelihood of a search; in particular, Native Americans remain the most likely
racial group to be subjected to either a low or a high discretion search, and African
Americans and Hispanics are slightly more likely to be searched than are Non-Hispanic
Whites. Asians/pacific islanders and East Indians are less likely to be searched than NonHispanic White, African American, Native American or Hispanic drivers.
As we concluded in earlier reports, however, we find no evidence that these
disparities at the bivariate level are the result of intentional or purposeful discrimination,
and thus we find no evidence of intentional “racial profiling” (evidence of purposeful or
intentional discrimination is generally the first step required by the federal courts when
attempting to prove racial discrimination under the Equal Protection clause of the
Fourteenth Amendment of the U.S. Constitution).

We come to this conclusion by

comparing the likelihoods of high discretion searches to low discretion searches, which
suggest that officers do not act differently based on race when they have higher levels of
discretion.

Moreover, the multivariate analysis results indicate that while race is

correlated with the both low and high discretion searches, there are multiple factors at

43

 play in what is most certainly a complex event. The predicted probabilities reported in
Table 23 indicate that of the driver characteristics, age and sex may have as important an
impact on the likelihood.
For instance, a Non-Hispanic White, 18-year old male has a probability of .004 of
being subjected to a high discretion search while an 18-year old Hispanic female has a
probability of .003, and a 50-year old African American or Hispanic female has a
probability .0007 of being subjected to a high discretion search. Moreover, the hit rate
analyses reported indicate that at least for Non-Hispanic Whites, African Americans,
Hispanics and Native Americans, the WSP achieves fairly efficient policing. In fact, the
hit rates for Non-Hispanic Whites, African Americans and Native Americans are exactly
the same for low discretion searches. The low hit rates for Asians/pacific islanders and
East Indians suggest that perhaps members of those groups are being searched at too high
of rates to achieve what is commonly regarded in law enforcement management as
“efficient policing.” These results are consistent with our earlier analysis, and are also
consistent with findings based on both quantitative and qualitative data.
Although we find no evidence of intentional discrimination or bad purpose, there
remains a statistical disparity at the bivariate level of analysis in the proportion of Native
Americans, blacks and Hispanics who are searched compared to whites (and of course a
disparity that favors Asians/pacific islanders and East Indians). We have been unable to
determine what might explain these disparities given the lack of any official WSP policy
to target minority drivers or other evidence of intentional discrimination or bad purpose.
We continue to believe that at least some of the disparity must be related to circumstances
or events that occur before the search and that are not, and perhaps cannot be, captured by

44

 these quantitative data. Because a majority of searches conducted are search incident to
arrest, the events that lead to the arrest might hold the key for at least part of the
explanation. In any event, the WSP may want to consider the matter further in order to
determine what may be the underlying cause of these disparities short of intentional
discrimination.

The Analysis of Use of Force for Evidence of Biased Policing
Introduction
For the first time over the five-year course research being done by researchers in
the Division of Governmental Studies and Services at Washington State University
racially-coded data on use of force was available for analysis with respect to biased
policing outcomes. Such data were made available for analysis for part of 2005, and very
importantly enhanced data for 2006-2007 – digital records which contain far more
information and more richly detailed accounts of uses of force associated with the threat
of use of tasers – permits an excellent preliminary analysis of patterns of use of force to
be presented in this report. During the Spring of 2007 one of the members of the WSU
DGSS research team was permitted to inspect five case folders (selected at random
among those with both minority and non-minority subjects) from which the data being
analyzed are extracted. This review of the original document file for each of the five
cases indicated that all of the informational elements found in the digital data given to the
WSU researchers were accurate representations of the original document files in
question.

45

 As with the other sections of this project final report, a major effort was extended
by the Washington State University research team to collect and organize as much data
as possible in the quite limited timeframe of the NHTSA grant, and then submit
preliminary findings in an end-of-project report with additional analyses being submitted
as Addenda at some later date.

More detailed analysis of the use of force data –

particularly with respect to the effects associated with the adoption of tasers as a form of
non-lethal intermediate element of force – will be presented in the form of an addendum
to this report.

The 2005 Use of Force Data
The 2005 use of force data provide a useful, but rather limited range of potential
for analysis. These data feature information on only the highest level of force employed,
and as a consequence of this limited amount of information on the use of force incident
the range of analyses possible is very limited. Insofar as some analysis can be done with
the 2005 data (number of cases =270, for the period 01/05-12/05) it is apparent that there
was no relationship between likelihood of experiencing more severity of force application
and the race/ethnicity of the suspect when only the most severe tool or method of
controlling behavior by the officer is recorded for our analysis.
Using the WSP use of force data provided for 2005, across the 270 incidents of
use of force for which digital records are available for analysis after appropriate coding
for the race/ethnicity of the subjects and officers involved there was no relationship
between race/ethnicity and level of severity of force employed as a last resort by the WSP
officer. The following results can be presented in this regard for this report:

46

 Severity of Force

Non-Minority

Minority

Low Level of Force

0

1

(0.00%)

(Verbal command)

Moderate Level of Force

91
(55.15%)

(OC/Chemical, Personal Weapon, Flashlight,

(1.52%)
34
(51.52%)

PR-24, Escorts, Counterpoint, Taser, Asp)

Intermediate Level of Force

71
(43.03%)

(Hair hold, Total Limb Control, Take Down)

High Level of Force (Lethal)

3
(1.82%)

(Shotgun/rifle, Vehicle, Handgun)

29
(43.94%)
2
(3.03%)

While these results are likely rather reassuring to the WSP administrative team and
Troopers alike, it needs to be pointed out that the limited nature of the 2005 data collected
for monitoring and subsequent analysis greatly limits the depth of analysis that can be
accomplished.
The 2006-7 Use of Force Data
The data collected on use of force since the beginning of 2006, after the
deployment of the taser as a major non-lethal tool of intermediate force, is much richer in
detail and more fully presents the context within which the officers’ use of force
decisions can be analyzed. Because of this richness and level of contextual detail on
officer actions and suspect behaviors, it is possible to come close to a replication of
research done on use of force by respected Criminal Justice researchers who have studied
this phenomenon carefully and commonly use a proportionality test to indicate the
presence or absence of correspondence between a suspect’s behaviors and an officer’s use
of force response.

47

 Such a test permits an evaluator to determine the proportion of cases of use of
force when there is appropriate proportionality between suspect behavior (e.g., passive
resistance, active resistance, treat of use of weapon, actual use of weapon, etc.) and
officer reaction (e.g., verbal command, OC spray, take down, display of weapon, etc.),
and the proportion of cases wherein there is EITHER less than and more than a
proportionate use of force response. In such an analysis, if the percentage of minorities
and non-minorities present in either case of “disproportionality” is large, it may be the
case that biased policing outcomes are in evidence. For example, if cases of less than
proportionate force being used feature higher percentages of non-minorities than
minorities, biased policing may be present. Similarly, if cases of more than proportionate
force being used feature higher percentages of minorities than non-minorities, biased
policing may be present.
The categories of use of force by the officer taken from the official use of force
records are arranged in ascending order of potential harm thusly, in accord with the
research literature in this area:
Verbal command or threat: Verbal Command, Handcuff only
Taser Display or deploy as treat of use
Restraint and control: Hair Hold, OC/Chemical
Pain compliance/takedown: Escorts, Counterpoint, Total Limb Control, Take Down
Intermediate Weapons: Personal Weapon, Flashlight, PR-24, Taser apply
Deadly force: Shotgun/Rifle, Vehicle, Handgun

The categories of arrestee conduct recorded in the use of force records are arranged in
ascending order of resistance/non-compliance as follows: Complaining but compliant,
passive resistance, active resistance, assaulting behavior, life-threatening behavior,
possession of knife, possession/use of a firearm. The following distribution of cases
indicates a high degree of proportionality [shaded cases] (78.1%).

48

 Most Harmful force
Subject
Highest
Level of
Action

Verbal
command
or threat

Taser
Display
or
deploy

Restraint
and
control

Pain
compliance/
takedown

Intermediate
Weapons

Deadly
force

8

1

1

2

0

0

12

Total

Complain
Passive
Resistance
Active
Resistance
Assaulting
Behavior
Life
Threatening

6

49

4

18

0

0

77

13

89

17

113

4

5

241

6

27

9

67

4

0

113

0

2

0

2

0

0

4

Had a Knife

0

2

0

2

0

0

4

Used or had a
Gun

0

2

2

1

3

6

14

Total

33

172

33

205

11

11

465

It is reassuring that the percentage of cases falling in the appropriate use of force given
the conduct of the arrestee, the important question to be addressed vis-à-vis racial
profiling or biased policing concerns is the ethnic/racial breakdown within the “less
force,” “commensurate force,” and “more force” categories. The following table sets
forth findings from the 2006-2007 WSP use of force dataset.

Force Factor Score
Less force
Commensurate force
More force
Total

Non-minority

Minority

X2

Total

41

14.00%

27

15.60%

68

14.60%

231

79.10%

132

76.30%

363

78.10%

20

6.80%

14

8.10%

34

7.30%

292

100%

173

100%

465

100%

.522(2)

Chi Square = .522 (2 degrees of freedom). Impact of minority/non-minority distinction is NOT
statistically significant

In the area of use of force, there is no evidence of systematic bias in the application force
vis-à-vis racial/ethnic minorities. Our preliminary analysis revealed that the introduction
49

 of the taser into the use of force situation has occasioned some dramatic shifts in the
frequency of use of some tools of control (e.g., OC spray). In our follow-up analyses to
be submitted as an addendum to this final report we will perform a detailed analysis of
these effects.
The final section of this report moves back to a fundamental question we
addressed at the outset – namely, the question of the most appropriate “denominator” for
assessing the presence of racial profiling. For the analysis of traffic stops, citations and
searches we have used four different estimators of the denominator in an attempt to
present as clear a picture as possible of the possible presence of biased policing. The
final section of this report sets forth a test of the proposition that one of these four
estimators is likely a very close surrogate measure of the composition of the driving
public on a particular roadway as validated in an observational study.

Preliminary Results of the Use of Observational Studies for
Denominator Assessment

50

 Introduction: The Search for a Reliable Denominator
A major issue for the assessment of the presence or absence of biased policing
relates to the question of the proper “denominator” to be used against which to compare
rates of stops, citations, and searches in any geographic area. Oftentimes the most readily
accessible data derived from the U.S. Census or from state-level estimated demographics
such as the OFM in Washington have little if any relevance for determining the
proportion of population of varying racial and ethnic backgrounds that might be present
on roadways of interest. The highways and/or roads in question in many cases serve as
pass-through lanes of travel for motorists who do not resemble the resident population of
the area of concern.
Consequently, a great deal of effort has been expended in recent years by a
number of scholars working in this area of research to make the argument that only
through direct field observation by two or more coders recording the racial and ethnic
composition of drivers on a particular roadway is it possible to establish an accurate
denominator estimate {Lamberth, John (1996), Revised Statistical Analysis of the
Incidence of Police Stops and Arrests of Black Drivers/Travelers on the New Jersey
Turnpike between Interchanges 1 and 3 from the Years 1988 through 1991. Report of
defendant’s expert in State v. Pedro Soto, 734 A2d 350 [N.J. Super. Ct. Law. Div. 1996]}.
Given the great expense associated with this difficult process of field observation, some
scholars have suggested that ACCIDENTS coded for race and ethnicity are an acceptable
accurate substitute for field observation-based studies {Smith, Michael R. and Geoffrey
P. Alpert (2003), “Searching for Direction: Courts, Social Science, and the Adjudication

51

 of Racial Profiling Claims,” Justice Quarterly 19: 673-703 and Smith, Michael R.
(2000), The Traffic Stop Practices of the Richmond, Virginia Police Department:
Final Report}. In this regard, it is argued that accidents are principally a random event,
and hence should affect all drivers relatively equally. It is the question of the reliability
of racially-coded accident data that is the subject of study in this aspect of the NHTSAfunded research carried out by the research team from the WSU Division of
Governmental Studies and Services.

Observation Studies Conducted with the Aid of Digital Photography
In previous research involving the systematic field observation of driver
characteristics a substantial number of observers working in teams and working multiple
shifts labored to collect observations and code those observations as they occurred. The
only formal record of that work, unfortunately, is the paper record prepared by each
coder, with a subsequent comparison of coder paper records being used to establish a
level of inter-coder reliability for the observations in question. Given the extremely high
personnel costs associated with this type of work, the norm is for only two coders to be
used on an observation (see: Steven K. Smith, Carol J. DeFrances and Carolyn C.
Williams, Assessing Measurement Techniques for Identifying Race, Ethnicity, and
Gender: Observation-Based Data Collection in Airports and at Immigration
Checkpoints. Washington, DC: U.S. Department of Justice, Bureau of Justice Statistics,
Jan. 2003)
The WSU Division of Governmental Studies and Services team attempted to raise
the standard for field observation research of this type by securing digital photography

52

 equipment and associated computer hardware and software that would permit the
accomplishment of these very important enhancements over previous work in this area:


Collecting large numbers of facial images of drivers



Using random selection of images to ensure unbiased choice of observations



Collecting data that constitute a record of observation which is permanent
and subject to re-analysis and replication by any interested party



Using multiple coders of diverse racial and ethnic backgrounds for the
coding of facial images



Collecting data in multiple locations (South King Co., Spokane, and Yakima
Valley) varying markedly in racial and ethnic composition

The availability of moderately-priced high speed cameras and high resolution lenses
which can be connected to computers to store large quantities of roadway digital images
made it possible test a new method of conducting field observations for this study. In
three diverse locations across the state the local WSP district office provided on-site
assistance in identifying safe and strategically positioned observational sites from which
a DGSS field observation team could collect facial images of vehicle operators in each
location. The WSP provided agency vans that could be adapted for the purpose of
providing an appropriate platform for the camera, computer equipment, and equipment
operators (two persons) needed for the collection of large numbers of on-site
photographically documented observations.
With the technical assistance of subcontractor Norman McDonald, an Information
Technology and Photographic Technology specialist, we were able to secure the use of
the necessary camera and computer hardware and software and receive training for DGSS

53

 researchers to collect driver facial images in our primary area of interest where our
previous research lead us to endeavor to “drill down” into traffic stop phenomena in this
area of the state – namely, the South King County area (APA 6) featuring a heavily
traveled route [Highway 99] just north of the Seattle/Tacoma International Airport
(SEATAC) where a highly diverse local population resides and much of the traffic on the
roadway is composed of racially and ethnically diverse persons. Having an accurate
denominator for assessing traffic stop data in this APA was essential to our “drill down”
work. As it turned out, the photographic equipment and computer software needed for
extracting and enlarging facial images from full intersection time-lapse images worked
even better than expected (i.e., yielded a rich number of useable facial images for
coding), and it was possible to increase the number of observation sites from the single
site in western Washington to two additional sites in eastern and central Washington to
test out the reliability of this method of research in both urban and rural settings.
Comparison of Observational Results with Accident Records on US 99 (APA 6)
The original purpose of conducting the observational study in APA 6 was to
determine as best as possible what the actual racial and ethnic make-up of the driving
public is on that highly traveled roadway. Since the incidence of traffic accidents on that
road provides a substantial record against which the observational data could be
compared, this location was our original choice for conducting an observational study. It
became apparent once in the field that the “bugs” in the new equipment and the learning
curve associated with using a new method of observation we anticipated with a first-time
use of an untried approach to data collection did not arise as feared. Given this good
fortune, it became possible to both code a large number of images on computer screens in

54

 Pullman using multiple coders AND replicate the use of the same method of digital data
collection in two additional research settings. Those two additional replications of the
use of the method, and the results obtained from those efforts, are described below in the
next section of this report. At this point it suffices to note that these two additional tests
of the digital photography field observation method add further to our confidence in this
approach to denominator estimation research.
The following summary statistics were derived from WSP traffic stop data for the
period November 1, 2005 through September 30, 2006, and reflect a total of 4,052 traffic
accidents to which the WSP responded in APA 6 in that period. Of those accidents, a
total of 3,019 featured racial/ethnic coding information.
Number

Percentage

Non-Minority Drivers

1,991

65.9%

Minority Drivers

1,028

34.1%

The comparable figures, derived from field observation digital photographs collected on
US Highway 99 in the Spring of 2007 (April 11-13) and coded by five racially and
ethnically diverse coders – one Native American, two blacks, two Anglos (three males
and two females) – represent a preliminary test of the utility of the foregoing accidentbased estimates of a denominator for APA 6. The figures reported here represent those
cases in which 4 out of 5 coders agree on the minority/non-minority coding of a digital
facial image [456 images of 692 coded (67%)] selected at random from the 6,198
intersection images recorded on US Highway 99 at the municipal boundary for Tukwilla.
Number

Percentage

Non-Minority Drivers

305

66.9%

Minority Drivers

151

33.1%

55

 As can be seen from the comparison of the two sets of figures, the estimation of rates of
minority and non-minority drivers from accidents would seen to be quite warranted. The
findings from the accident records and the results of the observational study are very
nearly identical. It is highly unlikely that such a close match in the results of the two
independent measures of the “denominator” would coincide so closely by chance.
Furthermore, a review of the preliminary analysis of findings from a replication of the
digital photography-based observational study process in the Spokane area would seem to
add further to the conclusion that accident rate data are likely to be a reliable surrogate
measure for the racial/ethnic composition of the driver population on a high volume
roadway.
Preliminary Results for Spokane
In the Spokane area the Spokane Police Department has been collecting accident
data coded for race and ethnicity since January of 2005, and those data are sent to DGSS
on a monthly basis for the compilation of a database to be used to compare against traffic
stop data being collected in mobile data terminals being installed in the agency’s patrol
vehicles.

As of 01/01/05, a total of 1,594 traffic accidents have been coded with

racial/ethnicity information, with 183 of those accidents occurring on Division Street (a
major North-South thoroughfare), all involving two drivers (n=366) and each driver
being coded for race, gender and ethnicity. The comparable figures for the Spokane
replication of the Seattle-area study are as follows:
Spokane PD Records for Division
Number
Non-Minority Drivers

351

56

Percentage
95.9%

 Minority Drivers

15

4.1 %

Spokane Digital Photography Observational Study (Summer 2007)
Number

Percentage

Non-Minority Drivers

443

97.1%

Minority Drivers

13

2.9%

This preliminary analysis of data collected in Spokane (4,658 intersection images from
time-lapse setting; 541 facial images “harvested” at random from those images collected
May 29-31, 2007) was conducted with four coders of diverse ethnic and racial
background. In all cases wherein three out of the four coders agree on minority/nonminority category assignment with a high rating of image clarity and high level of
confidence in judgment (n=456), it is evident that racially and ethnically coded accident
data are very likely an excellent source of information for the estimation of
denominators for assessing racial profiling phenomena. Of course, these findings are
best considered preliminary; both more coding work and more thoroughgoing analytical
work will be done to prepare an Addendum to this report on the use of accidents for
denominator estimation at a near future date.
Yakima Valley Replication
Both the “Westside” (high-concentration minority population area) and “Eastside”
(low rate of minority population) locations involved URBAN settings with large, multilane intersections featuring high volumes of traffic being available for setting up strategic
observation sites. Our interest in racial profiling phenomena is not restricted to urban
areas, however, and it is very important to know whether the digital photograph

57

 observation methodology can be employed in rural settings as well where moving
vehicles are the subjects of study.
In accord with this need to “test the limits” of the digital photography observation
methodology with moving vehicles we solicited and received support from the Union
Gap office of the WSP to collect moving vehicle observation data on US Highway 82.
This is a heavily traveled roadway where WSP records indicate a high proportion of
Latino drivers being involved in traffic stops, citations, and accidents. Upon preliminary
analysis, it is clear that the technology we secured for this type of field observation work
DOES WORK with moving traffic in rural areas where manual activation of the shutter
release is required. A total of 723 useable facial images were extracted from 2,914
images collected in two days (July 14 and 15, 2007) of field observation of moving
vehicles. At this point no coding has been done on these images, but such an analysis
will be done and reported in an addendum to this report.
Conclusion
It has been argued by many law enforcement agencies that the collection of
racially coded data on traffic stops is not advisable because the presence or absence of
racial profiling can only be established if an accurate “denominator” is available. It is
argued further that since the determination of such an accurate denominator requires very
expensive observational studies, the desire to collect racial profiling-relevant traffic stop
data is severely tempered by the high costs associated with observational studies. Our
research, reported at a preliminary stage of analysis, suggests very strongly that two facts
need to be taken into consideration on the question of the advisability of collecting
racially coded traffic stop data:

58

 1. The cost of observational studies can be greatly reduced and methodologically
enhanced by using digital photography
2. The use of racially-coded accident data as a surrogate denominator is likely an
acceptable alternative to the collection of observational where the latter is
cost-prohibitive for a police jurisdiction
Given the short period of time available for the 2007 study – involving a very large
(n=11,000+) multi-wave statewide mail survey, the analysis of 500,000+ traffic stops, and
the analysis of use of force data – it was decided that the major effort would be devoted
to the collection of as much pertinent data as possible, and the preparation of a set of
preliminary findings for the end-of-project report. The findings reported here are,
therefore, best considered preliminary and somewhat provisional. Additional work will
be undertaken with these data in the months ahead, and Addenda to this section and some
of the other sections of the report (e.g., use of force and taser implementation) will be
submitted to the WSP as they are developed.

59