ArticlePDF Available

How Suspect Race Affects Police Use of Force in an Interaction Over Time



Although studies often find racial disparities in policing outcomes, less is known about how suspect race biases police interactions as they unfold. This study examines what is differentially occurring during police–suspect interactions for White, Black, and Latino suspects across time. It is hypothesized that racial bias may be more evident earlier in interactions, when less information about the situation is available. One hundred thirty-nine (62 White, 42 Black, and 35 Latino) use-of-force case files and associated written narratives from a medium to large size urban police department in the United States were analyzed. Trained coders broke down the interaction narratives into discrete “sequences,” or dyadic action–reaction steps involving a suspect action (level of resistance) and an officer response (level of force). A linear mixed-effects model was run on amount of police use of force by suspect race and time, with suspect resistance and suspect actions toward third-party/self as controls. Results demonstrated that Black and Latino suspects receive more force in the beginning stages of the interaction, whereas Whites escalated in level of force faster after initial levels. By breaking down police–suspect interactions into discrete sequences, the current study reveals a better understanding of when bias originates in police use of force and informs how to focus policing interventions.
Law and Human Behavior
How Suspect Race Affects Police Use of Force in an
Interaction Over Time
Kimberly Barsamian Kahn, Joel S. Steele, Jean M. McMahon, and Greg Stewart
Online First Publication, October 20, 2016.
Kahn, K. B., Steele, J. S., McMahon, J. M., & Stewart, G. (2016, October 20). How Suspect Race
Affects Police Use of Force in an Interaction Over Time. Law and Human Behavior. Advance
online publication.
How Suspect Race Affects Police Use of Force in an Interaction Over Time
Kimberly Barsamian Kahn, Joel S. Steele, Jean M. McMahon, and Greg Stewart
Portland State University
Although studies often find racial disparities in policing outcomes, less is known about how suspect race
biases police interactions as they unfold. This study examines what is differentially occurring during
police–suspect interactions for White, Black, and Latino suspects across time. It is hypothesized that
racial bias may be more evident earlier in interactions, when less information about the situation is
available. One hundred thirty-nine (62 White, 42 Black, and 35 Latino) use-of-force case files and
associated written narratives from a medium to large size urban police department in the United States
were analyzed. Trained coders broke down the interaction narratives into discrete “sequences,” or dyadic
action–reaction steps involving a suspect action (level of resistance) and an officer response (level of
force). A linear mixed-effects model was run on amount of police use of force by suspect race and time,
with suspect resistance and suspect actions toward third-party/self as controls. Results demonstrated that
Black and Latino suspects receive more force in the beginning stages of the interaction, whereas Whites
escalated in level of force faster after initial levels. By breaking down police–suspect interactions into
discrete sequences, the current study reveals a better understanding of when bias originates in police use
of force and informs how to focus policing interventions.
Keywords: suspect race, police, use of force, racial bias, time
Although not a new phenomenon, recent police shootings of
racial minorities—like that of Michael Brown in Ferguson, Mis-
souri, in 2014 have produced increased calls for police reform
and reducing racial bias in policing (e.g., Kahn & Martin, 2016).
The President’s Task Force on 21st Century Policing (2015)
echoed the necessity to improve police trust with racial minorities
as one of its central findings. Given heightened racial tensions and
the life-or-death consequences involved, understanding and coun-
teracting racial bias in policing is of paramount importance. Al-
though empirical studies have thoroughly investigated racial dis-
parities on policing outcomes, less is known about how suspect
race biases police interactions as they unfold. The current study
investigates how suspect race influences use of force in police–
suspect interactions over time, thereby providing a more nuanced
look at how and when bias originates in policing.
Suspect race affects various outcomes in policing and the
broader criminal justice system, impacting interactions with po-
lice, arrest rates, and criminal sentencing (Kahn & Martin, 2016;
Kennedy, 1998; Walker, Spohn, & DeLone, 1996). Racial minor-
ities, and particularly Black males, are often found to have dis-
proportionate contact with the police (Brunson & Miller, 2006;
Skogan, 2005), including disproportionate pedestrian stops (Gelman,
Fagan, & Kiss, 2007; Ridgeway, 2007) and traffic stops (Cordner,
Williams, & Zuniga, 2000; Novak, 2004). Although results can be
mixed (e.g., Engel, Sobol, & Worden, 2000; Garner, Maxwell, &
Heraux, 2002), suspect race has been shown to affect the applica-
tion of force at lethal and nonlethal levels (Fridell & Lim, 2016;
Gau, Mosher, & Pratt, 2010; Goff, Lloyd, Geller, Raphael, &
Glaser, 2016; D. A. Smith, 1986; Terrill & Mastrofski, 2002;
Worden, 1995). Recent efforts to systematically track deadly force
and race find similar racial disparities (Gabrielson, Grochowski
Jones, & Sagara, 2014; Washington Post, 2016). These observed
racial disparities in policing point to the need to better understand
the source of these outcomes, whether they be primarily caused by
nonracial situational factors (e.g., neighborhood characteristics) or
officer racial attitudes (Goff & Kahn, 2012; Kahn & Martin, 2016).
Research on police–suspect interactions provides important in-
sights as to why and how police force is applied. Sykes and Clark’s
(1975) deference exchange theory has been used to understand the
dynamics of police–suspect interactions. Such interactions are
marked by unique characteristics between officers and suspects,
including an asymmetrical power difference in favor of the officer
and the expectation that citizens defer to police authority. When
this deference is not met, police may employ force to reestablish
the power dynamic of the relationship. Thus, suspect resistance is
often closely tied to the application of force (Alpert & Dunham,
1997). Perceived noncompliance may be exacerbated depending
on the race of the suspect and the perceived status of the police
officer, which can lead the encounter to escalate in severity (Ka-
duce & Greenleaf, 2000). Relatedly, perceived threat to the officer,
which may also vary by suspect race, is often cited to justify force
application (Wallentine, 2009).
The application of police force is often studied using police
use-of-force reports (Hickman & Atherley, 2012; Hickman, Ather-
ley, Lowery, & Alpert, 2015; MacDonald, Manz, Alpert, & Dun-
ham, 2003; McLaughlin, 1992; Stewart, 2013; Terrill, 2005; Ter-
Kimberly Barsamian Kahn, Joel S. Steele, and Jean M. McMahon,
Department of Psychology, Portland State University; Greg Stewart, De-
partment of Criminology and Criminal Justice, Portland State University.
We thank John Eubanks for his assistance and contributions to this
Correspondence concerning this article should be addressed to Kimberly
Barsamian Kahn, Department of Psychology, Portland State University,
P.O. Box 751, Portland, OR 97207. E-mail:
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
Law and Human Behavior © 2016 American Psychological Association
2016, Vol. 40, No. 6, 000 0147-7307/16/$12.00
rill & Paoline, 2012). Such reports may include check-box
outcomes detailing the level of force employed and/or narrative
reports by the officer stating what occurred that led to force
application (Terrill, 2001). Although primarily representing the
officer perspective, reports go through various layers of review for
accuracy (Terrill, 2001). Force can be conceptualized as a dichot-
omous outcome (application of force vs. not) or a continuum of
force, ranging from low to higher levels of force (Alpert &
Dunham, 1997; Garner, Schade, Hepburn, & Buchanan, 1995;
Terrill, 2003), and can be both verbal and physical in nature
(Klinger, 1995). On average, applied force is most often on the
lower end of the spectrum (Garner et al., 1995; Klinger, 1995;
Terrill, 2003), and deadly force is a rare event (Hickman, Piquero,
& Garner, 2008; Terrill, 2001).
Delving into the microprocesses that occur in these interactions
reveals important insights into how use of force is deployed. Using
this methodology, force encounters are sequenced by coding for
officer-force/suspect-resistance dyadic steps within an interaction.
That is, each act of suspect resistance is paired with the corre-
sponding officer action to create a step or sequence in the inter-
action (Hickman & Atherley, 2012; Hickman et al., 2015; Stewart,
2013; Terrill, 2003, 2005). Within these behavioral sequences,
there is a necessity of viewing force application in light of suspect
resistance throughout the dyadic interaction, as the two are closely
interconnected (Alpert & Dunham, 2004; Atherley & Hickman,
2014; Hickman & Atherley, 2012; Hickman et al., 2015; Terrill,
2003). Using this transactional approach, increases in suspect
resistance are often found to correspond with increases in police
force, and there is also evidence for the opposite directional
relationship, with police force leading to increased suspect resis-
tance (Terrill, 2003). Multiple acts of force and suspect resistance
can occur at varying levels within an interaction (Hickman &
Atherley, 2012; Terrill, 2003, 2005). What occurs early in an
encounter often sets the stage for how the remainder of the en-
counter will play out, such that early applications of force may lead
to later increases in officer use of force and/or suspect resistance
(Bayley, 1986; Worden, 1995).
Research has yet to examine what occurs differentially at a
micro, sequenced interaction level based on suspect race. Social
psychological research on how stereotypes and racial bias impact
interactions may shed further light on how suspect race may
influence police application of force across time in interactions.
Within the broader society and in the criminal justice context in
particular, Blacks and Latinos are stereotyped as criminals, aggres-
sive, and dangerous (Devine & Baker, 1991; Devine & Elliot,
1995; Madon et al., 2001; Niemann, Jennings, Rozelle, Baxter, &
Sullivan, 1994; Unnever & Cullen, 2012). These stereotypes and
beliefs may be held at either the explicit and conscious or implicit
and subconscious level, thereby not requiring intention to affect
behavior (Greenwald & Banaji, 1995; Greenwald, Poehlman, Uhl-
mann, & Banaji, 2009). Police officers are also aware of and may
hold these racial stereotypes, which can influence their interactions
with racial minority citizens and decisions to use force (Correll et
al., 2007; Eberhardt, Goff, Purdie, & Davies, 2004; Goff, Jackson,
Di Leone, Culotta, & DiTomasso, 2014; Plant & Peruche, 2005).
Racial stereotypes are often automatically activated in the per-
ceiver’s mind when they encounter an outgroup member, and
create expectations in the mind of the perceiver (Devine, 1989;
Hilton & von Hippel, 1996; Macrae, Bodenhausen, Milne, &
Jetten, 1994). Once activated, racial stereotypes serve as a lens
through which subsequent behavior is interpreted and shape per-
ceptions to be consistent with the stereotype (Banaji & Greenwald,
1995; Hamilton, Sherman, & Ruvolo, 1990; E. R. Smith, Stewart,
& Buttram, 1992; Stone, Perry, & Darley, 1997; von Hippel,
Sekaquaptewa, & Vargas, 1997). Stereotype application is most
powerful under conditions of ambiguity, scarce resources, anxiety,
and less information, as they serve to fill in the interpretation of
ambiguous actions (Bodenhausen, 1988; Bodenhausen & Wyer,
1985; Fiske, 1998; Hilton & von Hippel, 1996; Wilder, 1993). As
many policing encounters might involve these situational charac-
teristics, stereotypes may function similarly with police officers as
they interact with a suspect. When police officers first interact with
a racial minority suspect, racial group stereotypes may be auto-
matically activated, which can alter the ways in which officers
interpret subsequent actions (Hilton & von Hippel, 1996). That is,
a police officer interacting with a Black citizen may perceive the
individual in light of the criminal stereotype, consciously or un-
consciously, which can enhance perceptions of aggression and
danger. Thus, their behavior becomes assimilated to the stereotype
(Bodenhausen, 1990; Hilton & von Hippel, 1996).
How and when might racial stereotypes affect a police–suspect
interaction over time in a sequenced interaction? As suggested by
social psychological research, the current study focuses on initial
levels of force and the rate of subsequent change or escalation of
use of force. When more individuating information is available—
specific information about the individual or context—reliance on
racial stereotypes can be reduced (Fiske & Neuberg, 1990; Lock-
sley, Borgida, Brekke, & Hepburn, 1980). This process suggests
that racial stereotypes might have their strongest impact at the
early stages of an interaction, when less information is available
and stereotypes have been automatically activated.
Alternatively, it is also possible that officers may explicitly
attempt to suppress racial stereotypes upon initially engaging with
racial minorities during an encounter. Suppression of stereotypes
may be effective initially, but can often produce rebound effects
(Follenfant & Ric, 2010; Macrae et al., 1994; Wenzlaff & Wegner,
2000). Rebound effects occur when, after initial suppression of a
stereotype, stereotype activation and application becomes even
greater than prior to suppression. This pattern might suggest less
biased behavior at the start of an interaction as officers are sup-
pressing the stereotype, but predict a greater spike later in accor-
dance with a rebound effect. On the other hand, if police officers
do not hold stereotypes or any such biases at either the implicit or
explicit levels, race of suspects should have no notable differences
on their behavior throughout an interaction. Finally, as predicted
by a counterbias perspective, fear of racial bias may ultimately
lead to the reverse application of racial stereotypes and less force
used against racial minorities compared with Whites (James, Vila,
& Daratha, 2013). These alternative hypotheses are also examined
in the current study.
Despite extensive research highlighting disparities in use-of-
force outcomes by race, less is known about the process through
which use of force is applied and how suspect race may influence
this application. The current study therefore examines what un-
folds differently in police/racial-minority-suspect interactions than
police/White-suspect interactions over time. By examining police
officer use-of-force case narratives, the use-of-force incident is
broken down into discrete sequences. We then analyze how these
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
incidents unfold over time based on suspect race, focusing on how
time affects level of force throughout the interaction and rate of
change across time. Drawing from the social psychology literature
on stereotyping and implicit bias, it is hypothesized that suspect
race will affect use of force earlier in the incidents, such that racial
minorities will receive higher levels of force initially compared
with Whites.
Use-of-Force Case Files
Use-of-force case files were selected from a sample of 212
available incidents occurring during 2012 from a medium to large
metropolitan police department on the West Coast. The cases are
drawn from a predominantly White city, with non-White racial
minorities making up a smaller percentage of the population. All
cases involved at least one officer using force at some point in the
interaction, generating the reporting narrative. Cases involving no
force were excluded, as no narrative exists in these files. From the
available data, we selected cases to include only male suspects
(e.g., Paoline & Terrill, 2005; Schuck, 2004) and oversampled
racial minority cases to have a sufficient sample to test the study
hypotheses. Suspect racial groups that had at least 10 cases during
the time period were included for comparison (resulting in the
exclusion of Asians and Native Americans). This left White,
Black, and Latino suspects as available racial groups to examine,
which are representative of the city demographics. Although all
cases involved some level of force, selected cases involved only
nonlethal force application, as lethal force requires a separate
reporting mechanism and procedure outside the access of the
study. This included nearly all cases and is consistent with studies
of national use-of-force application that demonstrate that the vast
majority of police force encounters involve nonlethal force (Ter-
rill, 2001).
We further selected only cases involving crime types in which
all racial groups were represented in the larger sample, in order to
control for and test possible crime type differences between races.
These crime types included simple assault (n66), disorderly
conduct (n24), drugs (n10), aggravated assault (n9),
vandalism (n7), and cases coded as other (n23), which most
often represented interactions concerning warrants, detox holds, or
mental health checks. Because we were interested in examining
how use of force progresses over time, only cases in which at least
two rounds of interaction between the officer and suspect were
included. This created a sample of 139 cases for analysis consist-
ing of 62 White, 42 Black, and 35 Latino suspects.
Use-of-Force Narrative Coding
Use-of-force case files were coded into discrete sequences, or
dyadic action–reaction steps involving a suspect action (level of
resistance) and an officer response (level of force). This dynamic
sequencing method has been used to investigate change over time
in police–suspect interactions (Atherley & Hickman, 2014; Hick-
man & Atherley, 2012; Hickman et al., 2015; Terrill, 2003, 2005),
and has been validated as reliable by Stewart (2013) and Hickman
et al. (2015). Following this methodology, a total of five exten-
sively trained research assistants served as coders of the use-of-
force files. Training consisted of learning police tactics and termi-
nology, reviewing materials, and guided practice with the developed
coding scheme. Research assistants averaged 20 hr of instructor-
based training. A Use of Force Analysis Codebook, which pro-
vided definitions and examples, served as a reference for the
coders (see Stewart, 2013, for more information). Coding was
done using structured coding sheets that broke down each se-
quence in the interaction and required research assistants to check
the appropriate box for each outcome. Coders rated the following
variables in each use-of-force narrative.
Use-of-force sequences/sequence number. For each case,
coders read the first-person narratives written by police officers
following their use of force in the field. Every discrete interaction
between the responding officer and the suspect constituted a single
“sequence,” or step in the interaction. Each sequence began with
the suspect’s level of resistance to the officer and ended with the
officer’s subsequent actions toward the suspect, coded as level of
force. For the purpose of determining the sequence of events, a
pair of coders would meet and agree on the number of officer and
suspect interaction sequences before independently coding each
Suspect resistance. Each sequence began with the suspect’s
level of resistance toward the responding officer, measured on a 0
(no resistance; verbal exchange)to6(use of lethal force) scale
(Hickman et al., 2015; Stewart, 2013). A code of 1 (verbal/passive
resistance) represented the suspects’ verbal or nonverbal refusal of
commands from the officer. A code of 2 (use of posture and verbal
threats) represented the suspect’s verbal threats or enacting a
threatening posture. A code of 3 (physical noncompliance) indi-
cated the suspects’ attempts to avoid custody by fleeing, tensing
muscles, or pulling away from the responding officer. A code of 4
denoted active physical resistance, such as wrestling with or
striking the officer, and a code of 5 indicated the use of a nonlethal
weapon, such as the throwing of a rock or a chair.
Officer actions. This measure indicates the amount of force
used by the responding officer during the sequence. These re-
sponses were coded on a scale ranging from 0 (presence; verbal
exchange)to6(use of lethal force; Hickman et al., 2015; Stewart,
2013). As noted earlier, no lethal force cases were included.
Lawful orders, such as instructing the suspect to “stop running” or
“show me your hands,” were coded as a 1. A code of 2 for light
contact was associated with the officer touching the suspect in a
manner not meant to inflict pain, as when applying handcuffs or
escorting the suspect to a police vehicle. The code of 3 denotes a
physical control tactic. These are actions that are meant to force
compliance and may cause injury, such as tackling or pushing a
suspect to the ground, using joint manipulations to cause pain, or
placing the suspect in a hobble. A code of 4 for advanced physical;
chemical includes the use of fist, knee, and elbow strikes, as well
as pepper spray. The use of nonlethal weapons such as a baton,
Taser, canine, or a beanbag gun were coded asa5(intermediate
weapon use).
Suspect actions toward third-party/self. In many use-of-
force incidents, the suspect posed a threat to his or her own safety
(as during a suicide attempt) or the safety of a third party (a
domestic partner or another officer). This variable indexed the
suspect’s actions toward a third party or themselves, and not
toward the responding officer. These actions were coded on a scale
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
from 0 (no resistance; not applicable)to6(used lethal force/
actions) that is identical to the measure for suspect resistance.
Coding reliability. The dynamic sequencing use-of-force
method has previously been validated with police use-of-force
case narratives and has been deemed to be reliable (Hickman et al.,
2015; Stewart, 2013). Assessing the reliability of the specific study
sample, after agreeing on the sequences in the narratives, coders
individually rated 50 reports, with each report coded by two
coders. Measuring agreement between the raters, Cohen’s kappa
(; Cohen, 1960) indicated moderate to substantial interrater reli-
ability for the following variables: suspect resistance, ␬⫽.62,
95% confidence interval [CI] [.53, .72], p.001; officer’s actions
toward the subject, ␬⫽.78, 95% CI [.67, .83], p.001; and
subject’s actions to third-party/self, ␬⫽.54, 95% CI [.42, .65],
p.001. Given the acceptable to high agreeability ratings, par-
ticularly on the study variables of interest (e.g., officer actions and
suspect resistance), one of the case codes were randomly selected
for inclusion in the final data set for analysis.
Descriptive Statistics and Sample Characteristics
Table 1 provides descriptive statistics for the study variables of
interest, including maximum level of police use of force, suspect
resistance, suspect actions toward third parties or self, and number
of sequences. Maximum levels of force were, on average, below
the midpoint of the 5-point scale, indicating low to medium levels
of force. Level of suspect resistance similarly reflected low to
medium levels. Number of sequences ranged from two to 15, with
the average number of sequences between six and seven. Suspect
actions toward third parties were also on the low end of the
spectrum overall. These levels are consistent with past studies on
suspect resistance, use of force, and sequencing (Atherley &
Hickman, 2014; Hickman & Atherley, 2012; Hickman et al., 2015;
Terrill, 2003, 2005).
Testing racial group differences on these variables, ANOVA
analyses found no significant differences between suspect race and
maximum use of force, F(2, 135) .94, p.39, suspect resis-
tance, F(2, 135) .46, p.64, suspect actions toward third-
parties/self, F(2, 135) .68, p.51, and number of sequences,
F(1, 135) .33, p.72. Additionally, no significant differences
were evident using multivariate tests for maximum reported levels
of use of force, suspect resistance, suspect actions toward third-
parties/self, and number of sequences by race, Wilk’s ␭⫽.98, F(8,
264) .41, p.92, nor were differences evident for mean levels
among the dependent variables by race, Wilk’s ␭⫽.94, F(8,
264) 1.12, p.35. Thus, in addition to having comparable
crimes represented among the racial groups, the sample can be
characterized as relatively matched in terms of officer force, sus-
pect resistance, third-party/self actions, and number of sequences
in an interaction by suspect race.
Linear Mixed-Effects Modeling
In order to test the study hypothesis regarding how suspect race
impacts police use-of-force interactions over time, a linear mixed-
effects model was employed. The model examines the main hy-
pothesis as to whether Blacks and Latinos receive more force early
on in their interactions with police relative to White suspects.
Models were fit using the mixed-effects framework to account for
the nested nature of these sequences as well as to accommodate the
unbalanced nature of the interaction data. All continuous variables
were centered at their respective mean levels. Time (sequence
number) was entered as a predictor to model the linear regression
in officer actions over the interaction. In addition, time, represent-
ing a curvilinear quadratic relationship, was included, as it signif-
icantly improved the overall fit of the model (LRatio 15.85,
degrees of freedom 3, p.001).
Practically, this term was
included to capture any noticeable changes, such as possible peaks
or relative dips, in the level of force used by an officer during an
interaction. Notably, the inclusion of the quadratic term changes
the usual interpretation of the linear term, from an index of the
expected increase or decrease in force per time step for the entire
interaction, to a measure of the rate of increase or decrease at a
specific time point along the arc of interactions. For these analyses,
we selected the first interaction as the time point of interest; thus,
the linear terms represent the rate of change in force expected
at the earliest point of the officer–suspect interaction. Theoret-
ically, this instantaneous rate of force increase or decrease
should capture the officers’ reflexive actions toward the sus-
Suspect race was dummy coded such that Whites served as the
reference group. Therefore, all estimates that include the terms
“Black” and “Latino” represent the difference, either an increase or
decrease, from the same estimates for Whites as the comparison
category. Control variables included the level of suspect resistance
and the level of suspect actions toward third-parties/self. Interac-
tion terms were added for the linear and quadratic relationship with
time, suspect resistance, and actions toward third-parties/self with
suspect race. Additionally, the interaction around time was cen-
tered at the first time point, such that estimated racial group
differences are relative to the use of force at the beginning of the
sequence of interactions. All models were fit using the nlme
package in R, and fit using restricted maximum likelihood.
Adding a cubic term to time did not improve the model fit, and thus
was not added in the final model (LRatio 6.39, df 3, p.09).
Table 1
Descriptive Statistics by Suspect Race
Variable Suspect race Mean SD Min Max
Use of force White 2.02 1.49 0 5
Black 1.87 1.35 0 5
Latino 1.92 1.31 0 5
Suspect resistance White 1.99 1.36 0 5
Black 1.98 1.30 0 4
Latino 2.12 1.22 0 5
Suspect actions toward
White 1.73 1.51 0 5
Black 1.50 1.47 0 4
Latino 1.92 1.46 0 5
Number of sequences White 7.02 3.15 2 15
Black 7.05 3.35 2 14
Latino 6.23 2.46 2 11
Note. SD standard deviation; Min minimum score; Max maxi-
mum score.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
Results for the final model of officer use of force over time with
relevant covariates are reported in Table 2.
Effect sizes and 95%
CIs (lower limit and upper limit) are also reported. The model-
based R
values were computed based on work from Nakagawa
and Schielzeth (2013) and Johnson (2014), and ranged from .37 for
the marginal R
to .43 for the conditional R
. The average between
these values represent approximately 40% of the overall variance
in office use of force being explained by our reported model.
Figure 1 visually displays the different model-based trajectories
of police use of force for Black, Latino, and White suspects over
time. Examining the data, the hypothesis that Blacks and Latinos
would be subjected to more police force early on in interactions is
confirmed. Compared with Whites, Blacks (␤⫽.48, SE .21,
p.03), and Latinos (␤⫽.53, SE .23, p.02), each received
around a half point more force in the beginning of the interaction,
controlling for suspect resistance and actions toward third-parties/
self. Given the increase in severity associated with each step on the
5 point use-of-force scale, the half-point increase represents a
substantial and meaningful difference.
When examining how the interactions subsequently unfolded
over time after the initial force levels, time is a significant predic-
tor of police use of force (␤⫽.40, SE .06, p.001). This
indicates that White suspects were escalating in force as time
progressed, receiving just under a half-point increase in force for
every additional officer–suspect sequence during their interactions
with police. This rate of escalation varied, however, for Blacks and
Latinos compared with Whites. As indicated by the Black Time
(␤⫽⫺.30, SE .11, p.005) and Latino Time interactions
(␤⫽⫺.43, SE .13, p.001), the amount of increase for Black
and Latino suspects differed significantly from Whites. Black and
Latino suspects experienced less of an increase in force across the
interaction, indicating that the escalation of force was not as
pronounced for Black and Latino suspects as it was for White
suspects when all of the other covariates are taken into account.
The greater escalation in force for Whites compared with Blacks
and Latinos, however, should be interpreted in light of the signif-
icant racial differences in force earlier on. Whites may escalate
faster because they start from a lower force level compared with
Blacks and Latinos, who do not have as much capacity to subse-
quently increase as time progresses compared with Whites because
of the higher initial starting point in force.
Further examining the use-of-force trajectory over time, the
quadratic time effect is significant in the model (␤⫽⫺.02, SE
.01, p.001), which specifies an initial increase and eventual
decline in use-of-force levels as time progresses across interactions
for White suspects. As with the linear effect, Blacks (␤⫽.02,
SE .01, p.049) and Latinos (␤⫽.04, SE .02, p.003)
Additional variables were added to the model to see if the overall
model fit was improved using likelihood ratio tests with and without the
new predictor. Adding crime types as individual predictors (LRatio 5.75,
p.33) and officer experience (LRatio 1.12, p.89) did not improve
the fit of the model and were not kept for simplicity. The model did not
converge when officer race (White vs. minority) was added because of too
few data points.
Table 2
Restricted Maximum Likelihood Estimates for the Final Model of Officer Use of Force
Variables Value Std. error Effect size Lower limit Upper limit df t value pvalue
Intercept 1.26 .13 562 9.56 .000
Time .40 .06 .50 .38 .62 562 6.57 .000
.02 .01 .12 .18 .06 562 4.01 .000
Suspect resistance .16 .07 .14 .02 .27 562 2.27 .023
Suspect actions to third-party/self .35 .06 .37 .24 .50 562 5.54 .000
Suspect race: Black .48 .21 .34 .04 .64 135 2.23 .027
Suspect race: Latino .53 .23 .38 .05 .70 135 2.31 .023
Time Black .30 .11 .34 .55 .13 562 2.79 .005
Time Latino .43 .13 .33 .56 .10 562 3.34 .001
Black .02 .01 .10 .00 .21 562 1.97 .049
Latino .04 .02 .22 .08 .37 562 2.96 .003
Suspect Resistance Black .22 .11 .20 .00 .40 562 1.95 .051
Suspect Resistance Latino .28 .13 .26 .03 .49 562 2.20 .028
Action to Third-Party/Self Black .16 .10 .17 .37 .03 562 1.70 .091
Action to Third-Party/Self Latino .24 .10 .26 .48 .05 562 2.34 .020
Note. Std. error standard error; df degrees of freedom
Figure 1. Patterns of racial group averages in use of force across time
during police–suspect interactions. Plots are based on the 50th percentile of
number of sequences. Early differences in amount of force are apparent,
with Blacks and Latinos experiencing higher levels of force on average
during the beginning of exchanges with officers. Also illustrated is the
appreciable increase in force for White suspects over time, which is
significantly higher than for Black and Latino suspects.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
differ from Whites in the strength of this trend. To understand
these trends more fully, we combine the estimates from Whites and
each of the other racial groups in order to determine group specific
patterns based on the model. This is because of the dummy coding
used in these analyses, which makes the estimates of the quadratic
terms for both Blacks and Latinos into estimated differences for
these terms relative to Whites. In particular, Black suspects show
no appreciable quadratic effect over time because their estimates
differ from Whites in the exact opposite direction by the same
magnitude, (i.e., Whites Blacks, ␤⫽⫺.02 .02 0). For
Latinos, the resulting trend is positive quadratic (i.e., Whites
Latinos, ␤⫽⫺.02 .04 .02), indicating an eventual increase
in force used. This difference in the quadratic trend again may
result from the fact that they both started at higher levels of initial
Although our primary hypothesis of interest was to examine
change in use of force over time, the covariates also indicate
overall racial differences in treatment. First is the effect of suspect
resistance on officer use of force, shown graphically in Figure 2.
For White suspects, a unit increase in resistance receives a small
additional increase of approximately one sixth of a point in force
(␤⫽.16, SE .07, p.02). Consequently, this effect is
amplified for both Black and Latino suspects. Black suspects
receive more than an additional one-fifth-unit increase in force
(␤⫽.22, SE .11, p.05), and Latinos (␤⫽.28, SE .13, p
.03) receive more than an additional one-fourth increase in force
over and above White suspects when they resist. This equates to a
total of a .38 for Blacks and a .44 increase for Latinos in force for
a one-unit increase in resistance, respectively.
Finally, Whites who posed a threat to third parties or themselves
were associated with an overall increase in force (␤⫽.35, SE
.06, p.001). Recall that these were threats posed to people other
than the officer, which could include other individuals present at
the scene or threats to themselves. Blacks and Latinos differed
from Whites, such that threats to third parties from Latino suspects
(␤⫽⫺.24, SE .10, p.02) and marginally from Black
suspects (␤⫽⫺.16, SE .10, p.09) were associated with less
of an increase in force compared with Whites.
Examining how police–suspect interactions unfold over time
adds a unique perspective as to when and how a suspect’s race
alters police behavior. By breaking down police–suspect interac-
tions into discrete sequences, the current study reveals a better
understanding of how race can impact police use of force. Using
real-world police suspect use-of-force data and controlling for
relevant suspect and case variables, results indicate that both Black
and Latino suspects received higher levels of police force earlier in
interactions, whereas White suspects escalated in force at a greater
rate after the initial force levels compared with racial minorities.
When considering why initial rates of force were higher for
Blacks and Latinos, we suggest that racial stereotypes may, at least
in part, play a role during these initial actions. Racial stereotypes
associating Blacks and Latinos with danger may bias perceptions
at the beginning stages of an interaction, making the suspects seem
more threatening or in need of force to control. Indeed, psycho-
logical literature has demonstrated that stereotypes bias percep-
tions by filling in ambiguous information to be in line with the
stereotype, which can ultimately affect decision making (e.g.,
Correll, Park, Judd, & Wittenbrink, 2002). These stereotypes may
be held at either the explicit or implicit level, beneath individuals’
awareness, yet still bias behavior (Greenwald et al., 2009). Thus,
it is not necessary that officers consciously endorse or believe in
the validity of these stereotypes for their actions to be influenced
in accordance with the stereotypes (Kahn & Martin, 2016). Be-
cause stereotypes are more likely to influence perceptions and
behaviors when the situation is unclear or ambiguous and officers
may be lacking individuating information (Bodenhausen, 1988;
Greenwald & Banaji, 1995; Hilton & von Hippel, 1996), officers
may be more susceptible to the influence of stereotypes at the
beginning of an interaction, which is consistent with the present
data. As more individualizing information and case-specific details
are obtained as the interaction progresses, reliance on global ste-
reotypes may subside.
Alternative ways in which stereotypes might have affected
police behavior were not supported in the current data. A stereo-
type suppression and subsequent rebound effect (Macrae et al.,
1994) would suggest lower levels of force earlier on for Blacks and
Latinos, and then a sharper increase later in the interaction,
whereas a counterbias perspective (James et al., 2013) would
suggest lower levels of force throughout the interaction for Blacks
and Latinos compared with Whites. Both patterns are not consis-
tent with the current data, instead suggesting that stereotypes are
more influential earlier in the interaction.
It is also possible that officers feel the need to take charge early
in an interaction with racial minorities, perhaps anticipating resis-
tance or noncompliance in the encounter. In line with Sykes and
Clark’s (1975) deference exchange theory, they may then start
with higher levels of force to proactively control the situation and
Figure 2. Effect size estimates for suspect resistance based on suspect
race on officer use of force. These estimates are conditional, meaning that
they are relative to all other terms in the final model. Because suspect race
was dummy coded, with White serving as the reference group, the error
bars depicted represent the 95% confidence interval (CI) for the difference
of each race compared with Whites. The CI estimates for suspect resistance
for Whites represent the hypothesis that this value is equal to zero, which
is why there are not error bars depicted on the graph.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
establish the power dynamic. Anticipated resistance may also
enhance the perception of threat that the officer feels, which again
biases the interaction toward more severe applications of force.
Racial stereotypes or implicit bias may similarly be the root of this
anticipation of resistance and experience of threat.
After the disparities in initial force levels, use of force escalated
faster for White compared with Black and Latino suspects. This
steeper trajectory may be the result of a few factors. Because Black
and Latino suspects started at higher level of force initially in the
interaction, there may have been less rate of change compared with
Whites since their force levels were already capped, suggesting a
ceiling effect. It may further indicate that when force is used with
White suspects, it is deemed as more necessary or appropriate to
control a situation, as more situation specific information is gath-
ered passed the initial entry.
It also may indicate that implicit bias and stereotyping, rather
than explicit bias, might be the cause of the racial disparities. As
previously stated, implicit bias and stereotypes may play a greater
role earlier in an encounter when officers have less available
information and must make quick decisions. As the encounter
progresses and more information becomes available, disparities in
force may dissipate as officers are guided by less biased individ-
uating information and engaging in conscious, controlled process-
ing (Devine, 1989), which is consistent with this study’s data. If
bias was explicit and overt, racial disparities in force may instead
be consistent throughout the interaction, or even be exacerbated as
the interaction continued. This pattern is consistent with notions
that much of contemporary bias is implicit in nature (Dovidio,
2001; Greenwald et al., 2009), with policing being no exception
(Kahn & Martin, 2016).
It is also important to consider that this data set, at the surface,
reveals little racial disparities when examined at the final level of
force outcome alone. That is, there were no racial disparities in
maximum or average level of force, number of sequences in the
interaction, suspect resistance, and suspect actions toward third-
parties/self. As described earlier, cases included all male suspects
and were comparable crime types across racial groups. These
similarities in outcomes might lead one to conclude that suspect
race has no effect on the interaction. However, examining the
interaction at the sequence level reveals a different layer of racial
disparities. Black and Latino suspects receiving force earlier in the
interaction has important implications, even if maximum levels of
force ultimately equate to comparable White suspects. Time is
important for officers to assess the situation, engage in de-
escalation strategies, and gather information about the suspect and
scene. These findings suggest that this discretion is not taken to the
same extent with racial minority suspects as it is with Whites. The
threshold with which to use force might be lower for Black and
Latino suspects. More time and leniency might be afforded to
Whites, only escalating in force once it is deemed more necessary,
when other options are exhausted, or the situation becomes more
This suggestion is consistent with signal detection theory find-
ings from the shooter bias literature, which examine the underlying
judgment differences that lead to racially biased shooting deci-
sions. Evidence from shooter bias shows that, during video simu-
lations, racial stereotypes influence decision making such that
unarmed Blacks are shot more often than unarmed Whites (Correll
et al., 2002), although the racially influenced response pattern is
more often found in response times than shooting decisions with
police samples (Correll et al., 2007). Signal detection analyses on
these results show that individuals adopt a more liberal shooting
criterion when deciding to shoot Blacks compared with Whites,
which leads to a pattern of increased false alarms and decreased
misses (Correll et al., 2002; Kahn & Davies, 2011). That is, they
require less certainty in judgment before deciding to shoot Black
suspects, producing the racially biased shooter bias patterns. In the
current study, less certainty may be needed to engage force with
Blacks and Latinos early in the interaction, whereas more clarity is
required for Whites, leading to the delayed onset for this group.
Further, one can extrapolate that higher levels of force may have
been less necessary ultimately for Blacks and Latinos, but they
were not given the same opportunity for de-escalation as Whites.
As these data highlight, differential treatment by race within an
interaction can be masked by focusing only on ultimate outcomes.
The current data speak to how force is applied in interactions in
which force is eventually used. It is not representative of all
police–suspect interactions, as the majority of police–suspect in-
teractions do not involve force (Terrill, 2001). It is also possible
that Blacks and Latinos are more likely to receive force compared
with no force, making them more likely to end up in our current
data set compared with Whites. This argument would be consistent
with the adoption of a lower threshold for force application with
Whites being more likely to not receive force at all during police
Racial disparities were also highlighted in officers’ reactions to
the level of suspects’ resistance. When Black and Latino suspects
resisted, they received significantly more force than when White
suspects resisted. In line with racial stereotypes of criminality
(Devine & Elliot, 1995; Madon et al., 2001), this pattern may be
because officers view resistance from Blacks and Latinos as more
threatening or dangerous, in need of greater control. They may
therefore respond with increased force to subdue the perceived
threat. Alternatively, because of racial bias, the level of resistance
may be used to “justify” greater force against Black and Latino
suspects as punishment for noncompliance. Further, noncompli-
ance may be perceived differently coming from a racial minority
suspect, given the sociocultural context regarding perceived racial
bias in policing and the national discussions concerning the shoot-
ings of unarmed racial minorities. It may engender historically
tense relationships or broader notions of noncompliance and defi-
ance of police by racial minorities (Kaduce & Greenleaf, 2000), of
which the corresponding officer response is to increase force. This
effect may be heightened when it is a White officer interacting
with racial minority suspects, thus increasing the perceived soci-
etal status and power differences between the officer and suspect.
Similar disparities were also noted in regard to suspects’ actions
toward third parties and themselves. This measure represents the
level of perceived threat that suspects enacted toward others pres-
ent at the scene or toward themselves, for example, through threats
of suicide. Although these threats were associated with an increase
of force for all suspects, there was less of an increase for Black and
Latino suspects. This might suggest that threats made by Whites
were seen as more in need of immediate forceful action to subdue,
whereas those made by Blacks and Latinos were perceived less so.
These threats may have also been made toward different racial
groups present at the scene, which may be perceived with varying
levels of necessity for control. Threats toward other racial minor-
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
ities, for example, may be judged as less in need of immediate
action. Harm directed toward White third parties, and White
women in particular, may be acted upon more quickly or forcefully
in order to more fully protect White victims (e.g., Davis, 1983;
McMahon & Kahn, 2016). Because we did not differentiate be-
tween threats to different types of third parties (e.g., another
officer, an innocent bystander, an intimate partner) or the self, and
the race and gender of the third party present was not available in
the narratives (e.g., threatening an unidentified neighbor), we are
unable to test this hypothesis. Further examination is needed to
differentiate these potential causes and disaggregate these effects.
The current study carries implications for improving police
training and policy. First, it informs when to focus interventions to
combat police bias in use of force. It suggests that police training
should put particular emphasis on the beginnings of interactions
when stereotypes and race can have a larger potential impact. The
initial assessment of a scene and engagement with a suspect set
the stage for how the encounter unfolds and the levels of force the
encounter ultimately employs. Having these initial assessments not
be influenced by suspect race will lead to more equitable and fair
interactions. Further, training should also focus on suspect resis-
tance and perceived threats in the situation, as racial disparities
were found across interactions on these variables. These outcomes
may also be rooted in racial stereotypes, and thus training which
focuses on the reduction of both implicit and explicit stereotypes
on police behavior may be fruitful here as well. Police use-of-force
policies can also influence the application of force (Terrill &
Paoline, 2013), so policies may be useful in promoting more
equitable interactions. Policies that emphasize the use of more
de-escalation techniques or gathering of individuating information
before force is employed, particularly for racial minorities, may be
particularly effective in light of the current study’s findings.
Many policing agencies have traditionally relied upon on a
use-of-force continuum (also called levels of control) to help guide
officers in the appropriate application of force (e.g., Aveni, 2003;
Terrill & Paoline, 2013). This system is a mechanical model of
force in which the officer’s actions are guided by the suspect’s
level of resistance, with deviations in severity being deemed ex-
cessive. However, increasingly, courts have required officers to
more closely examine the “totality of the circumstance” in the
application of force (Graham v. Connor, 1989), which encourages
a system focused on the perceived threat posed to the officer
(Wallentine, 2009). With this change in emphasis away from
objective standards and guidelines to a more open or fluid defini-
tion, officers can use more force relative to resistance if they
perceived the suspect as being more of a threat. Reliance on
stereotypes or implicit bias may lead the officer to perceive more
threat initially in an interaction with racial minorities, above and
beyond resistance, which is then used to justify subsequent appli-
cation of force. Police standards that only emphasize perceived
threat as justification for force may therefore, unintendedly, exac-
erbate racial disparities in police use of force.
There are limitations that should be noted when considering the
findings from this study. As is common in criminal justice research
on police use of force (e.g., Hickman et al., 2015; Terrill, 2001),
the study used police use-of-force case narratives written by offi-
cers after a use-of-force incident to study the interactions. These
narratives are supposed to be an objective assessment of the scene,
with sufficient detail to be used as evidence should a case reach
court. They are often verified by other officers and supervisors for
correctness (Terrill, 2001). However, despite these protections,
they still primarily reflect the officer’s perspective of what hap-
pened in the situation, which might vary from the suspect’s inter-
pretation or from objective video evidence. Given that officer bias
might influence their narratives of the interaction in a way that
protects themselves or justifies their actions, that racial disparities
were still found is revealing and perhaps underestimates the
strength of the effects found. With the increased national push for
police to wear body cameras, future research might consider using
body camera recordings of use-of-force incidents to replicate these
findings. Other research may examine how closely these police
narratives match video recordings of incidents.
The current study is also limited in that its data are pulled from
one medium to large sized urban West Coast city in the United
States, which is not necessarily representative of all cities, and also
reflect a relatively small sample size because of the time-intensive
nature of the data coding. The city from which the police reports
were drawn is predominately White, suggesting limited generaliz-
ability in locations where the majority of denizens are racial
minorities. Therefore, caution should be taken in generalizing
these results across other cities in the United States. Further,
because of the exclusion of racial groups that did not have at least
10 suspects during the coded time period, we were not able to
examine any cases involving Asian or Native American suspects,
limiting generalizability to other racial minorities. Given these
limitations, it is suggested that this procedure be replicated in
different contexts and departments to gain generalizability. Fi-
nally, the small sample size necessitated analysis of only male
suspects. Disparities may also exist when examining suspect gen-
der in how an interaction unfolds, which could provide important
information. Further, suspect race and gender may also interact to
produce different patterns of interactions in unique ways, which
should not be discounted (Goff & Kahn, 2013).
Results from the current study highlight how suspect race dif-
ferentially changes and shapes an interaction between law enforce-
ment officers and suspects. Only by systematically understanding
when and how race pervades each aspect of a police–suspect
interaction will ways to counter its effects be truly possible.
Alpert, G. P., & Dunham, R. G. (1997). Policing urban America. Prospect
Heights, IL: Waveland Press.
Alpert, G. P., & Dunham, R. G. (2004). Understanding police use of force:
Officers, suspects, and reciprocity. New York, NY: Cambridge Univer-
sity Press.
Atherley, L. T., & Hickman, M. J. (2014). Controlling use of force:
Identifying police use of excessive force through analysis of adminis-
trative records. Policing: An International Journal of Police Strategies
& Management, 8, 123–134.
Aveni, T. J. (2003). Force continuum conundrum. Law and Order, 51,
74 –77.
Banaji, M. R., & Greenwald, A. G. (1995). Implicit gender stereotyping in
judgments of fame. Journal of Personality and Social Psychology, 68,
Bayley, D. H. (1986). The tactical choices of police patrol officers. Journal
of Criminal Justice, 14, 329 –348.
Bodenhausen, G. V. (1988). Stereotypic biases in social decision making
and memory: Testing process models of stereotype use. Journal of
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
Personality and Social Psychology, 55, 726 –737.
Bodenhausen, G. V. (1990). Stereotypes as judgmental heuristics: Evi-
dence of circadian variations in discrimination. Psychological Science,
1, 319 –322.
Bodenhausen, G. V., & Wyer, R. S., Jr. (1985). Effects of stereotypes on
decision making and information-processing strategies. Journal of Per-
sonality and Social Psychology, 48, 267–282.
Brunson, R. K., & Miller, J. (2006). Young black men and urban policing
in the United States. British Journal of Criminology, 46, 613– 640.
Cohen, J. (1960). A coefficient of agreement for nominal scales. Educa-
tional and Psychological Measurement, 20, 37– 46.
Cordner, G., Williams, B., & Zuniga, M. (2000). Vehicle stops for the year
2000: Executive summary. San Diego, CA: San Diego Police Depart-
Correll, J., Park, B., Judd, C. M., & Wittenbrink, B. (2002). The police
officer’s dilemma: Using ethnicity to disambiguate potentially threaten-
ing individuals. Journal of Personality and Social Psychology, 83,
1314 –1329.
Correll, J., Park, B., Judd, C. M., Wittenbrink, B., Sadler, M. S., & Keesee,
T. (2007). Across the thin blue line: Police officers and racial bias in the
decision to shoot. Journal of Personality and Social Psychology, 92,
1006 –1023.
Davis, A. Y. (1983). Women, race, and class. New York, NY: Vintage.
Devine, P. G. (1989). Stereotypes and prejudice: Their automatic and
controlled components. Journal of Personality and Social Psychology,
56, 5–18.
Devine, P. G., & Baker, S. M. (1991). Measurement of racial stereotype
subtyping. Personality and Social Psychology Bulletin, 17, 44 –50.
Devine, P. G., & Elliot, A. J. (1995). Are racial stereotypes really fading?
The Princeton trilogy revisited. Personality and Social Psychology Bul-
letin, 21, 1139 –1150.
Dovidio, J. F. (2001). On the nature of contemporary prejudice: The third
wave. Journal of Social Issues, 57, 829 – 849.
Eberhardt, J. L., Goff, P. A., Purdie, V. J., & Davies, P. G. (2004). Seeing
black: Race, crime, and visual processing. Journal of Personality and
Social Psychology, 87, 876 – 893.
Engel, R. S., Sobol, J. J., & Worden, R. E. (2000). Further exploration of
the demeanor hypothesis: The interaction effects of suspects’ character-
istics and demeanor on police behavior. Justice Quarterly, 17, 235–258.
Fiske, S. T. (1998). Stereotyping, prejudice, and discrimination. In T.
Gilbert & S. T. Fiske (Eds.), Handbook of social psychology (4th ed.,
Vol. 2, pp. 357– 411). Boston, MA: McGraw-Hill.
Fiske, S. T., & Neuberg, S. L. (1990). A continuum of impression forma-
tion, from category-based to individuating processes. Advances in Ex-
perimental Social Psychology, 23, 1–74.
Follenfant, A., & Ric, F. (2010). Behavioral rebound following stereotype
suppression. European Journal of Social Psychology, 40, 774 –782.
Fridell, L., & Lim, H. (2016). Assessing the racial aspects of police force
using the implicit-and counter-bias perspectives. Journal of Criminal
Justice, 44, 36 – 48.
Gabrielson, R., Grochowski Jones, R., & Sagara, E. (2014, October 10).
Deadly force, in black and white. Pro Publica. Retrieved from https://
Garner, J. H., Maxwell, C. D., & Heraux, C. G. (2002). Characteristics
associated with the prevalence and severity of force used by the police.
Justice Quarterly, 19, 705–746.
Garner, J. H., Schade, T., Hepburn, J., & Buchanan, J. (1995). Measuring
the continuum of force used by and against the police. Criminal Justice
Review, 20, 146 –168.
Gau, J. M., Mosher, C., & Pratt, T. C. (2010). An inquiry into the impact
of suspect race on police use of Tasers. Police Quarterly, 13, 27– 48.
Gelman, A., Fagan, J., & Kiss, A. (2007). An analysis of the New York
City police department’s “stop-and-frisk” policy in the context of claims
of racial bias. Journal of the American Statistical Association, 102,
813– 823.
Goff, P. A., Jackson, M. C., Di Leone, B. A. L., Culotta, C. M., &
DiTomasso, N. A. (2014). The essence of innocence: Consequences of
dehumanizing Black children. Journal of Personality and Social Psy-
chology, 106, 526 –545.
Goff, P. A., & Kahn, K. B. (2012). Racial bias in policing: Why we know
less than we should. Social Issues and Policy Review, 6, 177–210.
Goff, P. A., & Kahn, K. B. (2013). How psychological science impedes
intersectional thinking. Du Bois Review: Social Science Research on
Race, 10, 365–384.
Goff, P. A., Lloyd, T., Geller, A., Raphael, S., & Glaser, J. (2016). Science
of justice: Race, arrests, and police use of force. Los Angeles, CA:
Center for Policing Equity, University of California.
Graham v. Connor, 490 U.S. 386 (1989)
Greenwald, A. G., & Banaji, M. R. (1995). Implicit social cognition:
Attitudes, self-esteem, and stereotypes. Psychological Review, 102,
4 –27.
Greenwald, A. G., Poehlman, T. A., Uhlmann, E. L., & Banaji, M. R.
(2009). Understanding and using the Implicit Association Test: III.
Meta-analysis of predictive validity. Journal of Personality and Social
Psychology, 97, 17– 41.
Hamilton, D. L., Sherman, S. J., & Ruvolo, C. M. (1990). Stereotype-based
expectancies: Effects on information processing and social behavior.
Journal of Social Issues, 46, 35– 60.
Hickman, M. J., & Atherley, L. T. (2012). Police use of force in Seattle,
January 2009 –March 2011. Seattle, WA: Northwest Justice Solutions.
Hickman, M. J., Atherley, L. T., Lowery, P. G., & Alpert, G. P. (2015).
Reliability of the force factor method in police use-of-force research.
Police Quarterly, 18, 368 –396.
Hickman, M. J., Piquero, A. R., & Garner, J. H. (2008). Toward a national
estimate of police use of nonlethal force. Criminology & Public Policy,
7, 563– 604.
Hilton, J. L., & von Hippel, W. (1996). Stereotypes. Annual Review of
Psychology, 47, 237–271.
James, L., Vila, B., & Daratha, K. (2013). Results from experimental trials
testing participant responses to White, Hispanic and Black suspects in
high-fidelity deadly force judgment and decision-making simulations.
Journal of Experimental Criminology, 9, 189 –212.
Johnson, P. C. D. (2014). Extension of Nakagawa & Schielzeth’s R
GLMM to random slopes models. Methods in Ecology and Evolution, 5,
944 –946.
Kaduce, L. L, & Greenleaf, R. G. (2000). Age and race deference reversals:
Extending Turk on police-citizen conflict. Journal of Research in Crime
and Delinquency, 37, 221–236.
Kahn, K. B., & Davies, P. G. (2011). Differentially dangerous? Phenotypic
racial stereotypicality increases implicit bias among ingroup and out-
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
group members. Group Processes & Intergroup Relations, 14, 569 –580.
Kahn, K. B., & Martin, K. D. (2016). Policing and race: Disparate treat-
ment, perceptions, and policy responses. Social Issues and Policy Re-
view, 10, 82–121.
Kennedy, R. (1998). Race, crime, and the law. New York, NY: Vintage
Klinger, D. A. (1995). The micro-structure of nonlethal force: Baseline
data from an observational study. Criminal Justice Review, 20, 169 –186.
Locksley, A., Borgida, E., Brekke, N., & Hepburn, C. (1980). Sex stereo-
types and social judgment. Journal of Personality and Social Psychol-
ogy, 39, 821– 831.
MacDonald, J. M., Manz, P. W., Alpert, G. P., & Dunham, R. G. (2003).
Police use of force: Examining the relationship between calls for service
and the balance of police force and suspect resistance. Journal of
Criminal Justice, 31, 119 –127.
Macrae, C. N., Bodenhausen, G. V., Milne, A. B., & Jetten, J. (1994). Out
of mind but back in sight: Stereotypes on the rebound. Journal of
Personality and Social Psychology, 67, 808 – 817.
Madon, S., Guyll, M., Aboufadel, K., Montiel, E., Smith, A., Palumbo, P.,
& Jussim, L. (2001). Ethnic and national stereotypes: The Princeton
trilogy revisited and revised. Personality and Social Psychology Bulle-
tin, 27, 996 –1010.
McLaughlin, V. (1992). Police and the use of force: The Savannah study.
Westport, CT: Praeger.
McMahon, J. M., & Kahn, K. B. (2016). Benevolent racism? The impact
of target race on ambivalent sexism. Group Processes & Intergroup
Relations, 19, 169 –183.
Nakagawa, S., & Schielzeth, H. (2013). A general and simple method for
obtaining R2 from generalized linear mixed-effects models. Methods in
Ecology and Evolution, 4, 133–142.
Niemann, Y. F., Jennings, L., Rozelle, R. M., Baxter, J. C., & Sullivan, E.
(1994). Use of free responses and cluster analysis to determine stereo-
types of eight groups. Personality and Social Psychology Bulletin, 20,
379 –390.
Novak, K. J. (2004). Disparity and racial profiling in traffic enforcement.
Police Quarterly, 7, 65–96.
Paoline, E. A., & Terrill, W. (2005). Women police officers and the use of
coercion. Women & Criminal Justice, 15, 97–119.
Plant, E. A., & Peruche, B. M. (2005). The consequences of race for police
officers’ responses to criminal suspects. Psychological Science, 16,
180 –183.
Post, W. (2016). Fatal force [Database]. Retrieved from https://www
President’s Task Force on 21st Century Policing. (2015). Final Report of
the President’s Task Force on 21st Century Policing. Washington, DC:
Office of Community Oriented Policing Services.
Ridgeway, G. (2007). Analysis of racial disparities in the New York Police
Department’s stop, question, and frisk practices. Santa Monica, CA:
Rand Corporation.
Schuck, A. M. (2004). The masking of racial and ethnic disparity in police
use of physical force: The effects of gender and custody status. Journal
of Criminal Justice, 32, 557–564.
Skogan, W. G. (2005). Citizen satisfaction with police encounters. Police
Quarterly, 8, 298 –321.
Smith, D. A. (1986). The neighborhood context of police behavior. Crime
and Justice, 8, 313–341.
Smith, E. R., Stewart, T. L., & Buttram, R. T. (1992). Inferring a trait from
a behavior has long-term, highly specific effects. Journal of Personality
and Social Psychology, 62, 753–759.
Stewart, G. (2013). A quantitative method for the analysis of constitutional
factors in police use of force (Unpublished master’s thesis). Portland
State University, Portland, OR.
Stone, J., Perry, W., & Darley, J. M. (1997). “White men can’t jump”:
Evidence for the perceptual confirmation of racial stereotypes following
a basketball game. Basic and Applied Social Psychology, 19, 291–306.
Sykes, R. E., & Clark, J. P. (1975). A theory of deference exchange in
police-civilian encounters. American Journal of Sociology, 81, 584 –
Terrill, W. (2001). Police coercion: Application of the force continuum.
New York, NY: LFB Scholarly Publishing.
Terrill, W. (2003). Police use of force and suspect resistance: The micro
process of the police-suspect encounter. Police Quarterly, 6, 51– 83.
Terrill, W. (2005). Police use of force: A transactional approach. Justice
Quarterly, 22, 107–138.
Terrill, W., & Mastrofski, S. D. (2002). Situational and officer-based
determinants of police coercion. Justice Quarterly, 19, 215–248. http://
Terrill, W., & Paoline, E. A., III. (2013). Examining less lethal force policy
and the force continuum results from a national use-of-force study.
Police Quarterly, 16, 38 – 65.
Terrill, W., & Paoline, E. A., III. (2012). Conducted energy devices
(CEDs) and citizen injuries: The shocking empirical reality. Justice
Quarterly, 29, 153–182.
Unnever, J. D., & Cullen, F. T. (2012). White perceptions of whether
African Americans and Hispanics are prone to violence and support for
the death penalty. Journal of Research in Crime and Delinquency, 49,
519 –544.
von Hippel, W., Sekaquaptewa, D., & Vargas, P. (1997). The linguistic
intergroup bias as an implicit indicator of prejudice. Journal of Exper-
imental Social Psychology, 33, 490 –509.
Walker, S., Spohn, C., & DeLone, M. (1996). The color of justice: Race,
ethnicity, and crime in America. Belmont, CA: Wadsworth.
Wallentine, K. (2009). The risky continuum: Abandoning the use of force
continuum to enhance risk management. Retrieved from http://www
Wenzlaff, R. M., & Wegner, D. M. (2000). Thought suppression. Annual
Review of Psychology, 51, 59 –91.
Wilder, D. A. (1993). The role of anxiety in facilitating stereotypic judg-
ments of outgroup behavior. In D. M. Mackie & D. L. Hamilton (1st
ed.), Affect, cognition, and stereotyping: Interactive processes in group
perception (pp. 87–109). San Diego, CA: Academic Press Inc. http://dx
Worden, R. E. (1995). The causes of police brutality: Theory and evidence
on police use of force. In W. A. Geller & H. Toch (Eds.), Police
violence: Understanding and controlling police abuse of force (pp.
23–51). New Haven, CT: Yale University Press.
Received April 25, 2016
Revision received July 29, 2016
Accepted August 23, 2016
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
... The expectation is that influence of external review whether by civilians or other thirdparty stakeholders will lower departments' propensities to report force. Moreover, given findings that race and ethnicity influence use of force encounters (Kahn, Steele, McMahon, & Stewart, 2016;Wright & Headley, 2020) this research expects the black officer-to-black citizen ratio and Hispanic officer-to-Hispanic citizen ratios to reduce departments' likelihoods of reporting force since force is more frequently used in encounters against members of these groups (Kahn et al., 2016). ...
... The expectation is that influence of external review whether by civilians or other thirdparty stakeholders will lower departments' propensities to report force. Moreover, given findings that race and ethnicity influence use of force encounters (Kahn, Steele, McMahon, & Stewart, 2016;Wright & Headley, 2020) this research expects the black officer-to-black citizen ratio and Hispanic officer-to-Hispanic citizen ratios to reduce departments' likelihoods of reporting force since force is more frequently used in encounters against members of these groups (Kahn et al., 2016). ...
Using data from the 2013 Law Enforcement Management and Administrative Statistics (LEMAS) survey, the current study provides a cross-sectional analysis of U.S. police departments’ reported use of force. The goal of this study was to examine the extent to which departments’ reported force counts were explained by rational bureaucratic and/or institutional theory. Given the stark variations in reported force counts, a hurdle model was used to examine the potential effects of the theories on departments’ likelihoods of reporting force and the frequency in which they reported it. The results highlighted the significance of both theories. In terms of rational bureaucratic theory, the results illustrate that the absence of a collective bargaining agreement and greater professionalism requirements reduced departments’ likelihoods of reporting force, while less restrictive administrative policies increased departments’ likelihoods of reporting force and the frequencies in which they reported it. In terms of institutional theory, the results revealed that black officer representation reduced both the likelihood of reporting force and the frequency of force reported. However, increases in jurisdictions’ population and crime rates, for the most part, increased force reports. Combined the theories explained over one-fifth of the variations in departments’ reported use of force for the observed year. The findings suggest that successful efforts to reduce force-related injuries and deaths should consider the contextual environments in which rules and regulations regarding force are made.
... Die Verursachung dieser KV durch Kontextinformationen ist nicht nur bei Polizistys zu beobachten Luck & Ford, 1998;Richter et al., 2018;Stein & Peelen, 2015). Im Bereich des polizeilichen Interaktionsverhaltens mit Bürgys kann es sich bei solchen um Informationen wie beispielsweise Vorstrafen handeln, die mit einem Bürgy in Verbindung gebracht werden (Stolzenberg et al., 2021), oder um die Kenntnis der Ethnie, die zu Vorurteilen und Stereotypen führt Kahn, Steele, et al., 2017;Zhang & Zhang, 2021). Auch der Name des Bürgys selbst kann auf eine bestimmte soziale Gruppe oder Minderheit hinweisen und damit Vorurteile und Stereotype aktivieren (Boettner & Schweitzer, 2020 ;End, 2017). ...
Full-text available
Kognitive Verzerrungen (KV) sind im polizeilichen Interaktionsverhalten ein Problem und stellen die Polizei vor große Herausforderungen. Diese Herausforderungen ergeben sich aus der Art der KV, ihren Quellen und Fehlannahmen, die Polizistys 1 in ihrem Denken über sie fehlleiten. Auf der Grundlage eines Rahmenmodells, das KV und Fehlannahmen über ihre Zuschreibungsroutinen zueinander in Beziehung setzt, argumentieren wir, dass diese Fehlannahmen Bemühungen zur Abschwächung von KV im Kontext polizeilichen Interaktionsverhaltens untergraben. Basierend auf einem systemischen Verständnis von KV und ihrer schädlichen Auswirkungen kommen wir zu dem Schluss, dass die Implementierung reflexiver Strukturen innerhalb der Polizei eine notwendige Voraussetzung dafür ist, externe Einflüsse effektiv zu reflektieren und die Entfaltung von KV und entsprechenden Fehlannahmen in einer selbstreferentiellen Schleife zu verhindern.
This study examines racial and ethnic disparities in the use of force by the Fairfax County, VA Police Department using post-Ferguson data and after the adoption of modern training and use of force policies on de-escalation. White civilians were overrepresented as subjects of force compared to a variety of benchmarks, while Hispanics were underrepresented. Black and Asian civilians also were overrepresented using some benchmarks but not others. Multivariate models revealed that Black civilians experienced more severe force, higher average force levels, and more total force than Whites. We discuss possible reasons for the overrepresentation of Blacks and Whites as subjects of force, as well as the disparities in force levels and the totality of force experienced by Black civilians. We conclude with suggestions for policy change, training, and deployment changes to help reduce the potential influence of unconscious bias in the application of force by the FCPD and other law enforcement agencies.
Police officers’ use of lethal force is scrutinized to determine the appropriateness of the decision to shoot. However, surprisingly little research has systematically examined the sequence of behaviors that leads to this decision. This study content analyzed coronial reports of 58 deaths by police shooting in Australia, coding behavioral and situational features. A novel sequence analysis method—the proximity coefficient—was used to determine the proximities of police and subject behaviors to interpret how police firearm presentation and discharge may be influenced by preceding actions. Theoretical implications and practical applications for preventing fatal outcomes are discussed.
Use of force incidents involving police officers and civilians are complex, multi-faceted, and interactional. Officer force and civilian resistance are frequently measured at their maximum level or in relation to one another. While this approach is informative, it does not fully reflect the complexity of these encounters that contain a series of sequential actions taken by both parties. These processes are difficult to capture using traditional data collection efforts as they require time consuming independent coding of each action. Using data from two municipal police agencies, this study outlines a methodology for unpacking these complex interactions and examines sequentially based incident characteristics and their relation to the highest levels of force and resistance. Results reveal the importance of total actions and the starting point of force and resistance, which has specific implications for our understanding of how these interactional encounters change over time and police use of force more broadly.
This study explores retrospective social media experiences of college students from the St. Louis area and their level of engagement with content on police use of force, using qualitative semi-structured interviews (N = 32). Participants were asked about their online experiences with content showing the police use of force, the extent of their engagement with the content, and the types of police force they approve of. The study revealed that most participants who use social media experienced emotions of sadness, anger, and frustration when they observed the police use of force online, making them question police legitimacy. Although a majority of participants acknowledged police use of force news on social media helped them form opinions about law enforcement, only a small portion chose to share this content on their social media platforms. This is possibly due to the fear of negative backlash and/or employment and other opportunities they may be denied. Use of force approval rationales also varied by race when participants imagined hypothetical instances. The findings have implications for theories of media and race relations.
Police officers are expected to serve and protect civilians. Yet cases of police officers harming, sometimes killing, people they are expected to serve and protect have received renewed attention in the United States. Notable cases have led to protests, calls for reforms, and disputes about roles of police officers. This study employs Applied Sociology approaches to questions around reforms of policing in the United States. This study examines reforms to legal protections police enjoy through use of force and qualified immunity laws. Analyzing data for the 50 U.S. states and Washington, DC, this study examines whether legal protections police officers possess are associated with excessive use of force against civilians. This study finds that civilian killings by police tend to be higher in states whose governments are characterized as beyond civilian control. Where Republicans control state government and police officers enjoy broad protections, police killings of civilians tend to be higher. This study finds that racial conflict may be a powerful factor related to police killings of civilians. Applied Sociology demonstrates that reforms should concentrate on not only legal protections police enjoy but also societal structures shaping their efforts to serve and protect civilians.
People often assume areas where black people live are dangerous, impoverished, and rundown, whereas they assume White areas to be safe, wealthy, and well-maintained (Bonam et al. in J Exp Psychol Gen 145(11):1561–1582, 2016). These space-focused racial stereotypes shape how people perceive, evaluate, and treat physical space, such as houses and neighborhoods. Further, people often associate specific types of spaces with certain races (e.g., inner-city is Black; suburb is White), making them racial exemplar spaces (Bonam, in Devaluing Black space: Black locations as targets of housing and environmental discrimination, Unpublished doctoral dissertation. Stanford University, Stanford, 2010). The present work expands insight into racialized physical space by showing how Black and White exemplar spaces and White space-focused racial stereotypes shape person-perception. Specifically, we report findings from a vignette study examining how “black” and “criminal” people perceived a target person. This target was black or white and was in a suburban or inner-city neighborhood. We also measured people’s generalized stereotypes about white areas. People thought of the Black versus White person as being more black, which in turn made the target seem more criminal. This relationship was stronger in a suburban versus inner-city neighborhood—likely because being in an inner-city area made the White target seem more black, whereas the black target seemed black no matter where he was. Additionally, the more people thought of white areas as generally safe, wealthy, and well-maintained, the more they criminalized the black—but not white—target in a suburban neighborhood. This study highlights the need to further explore how racialized physical space shapes social perception, and it provides insight into the criminalization and policing of black bodies in white spaces.
American society and culture functions in a way that makes reflecting on our own biases uncomfortable. While those of us who work with people seek to be unbiased, we are inevitably susceptible to biases, because we are all human. It is essential for us to examine the influence of racism and implicit bias on our own lives and work. In this chapter, issues of identity, intersectionality, cultural humility, and minority stress will be explored. Clinical vignettes will be used to challenge readers to reflect on their own prejudices, including those may seem unconscious or subtle, and how they influence their interactions with young people and their interpretations of youths' behavior. The chapter also provides a framework we can use as the foundation for constructive open dialogue with our colleagues about all our thinking and behavior. By doing so, we become able to better serve those we work with, in addition to furthering our own growth as individuals.
Full-text available
Two studies investigated whether benevolent sexism is differentially applied based on a woman’s race. Study 1 demonstrated that participants expressed more benevolent sexism to White females than Black females when given no other information besides race. Study 2 introduced positive (chaste) and negative (promiscuous) sexually subtyped behaviors in addition to female race. Under these conditions, participants directed more benevolent sexism at chaste Black women rather than chaste White women, consistent with shifting standards theory. Despite receiving more benevolent sexism, chaste Black women did not receive more positive evaluations overall. Across both studies, expressions of hostile sexism did not differ by race. Results suggest that race may function as a subtype to elicit benevolent sexism contingent on behavior. Black women who follow traditional gender norms may be overcompensated for their conformity with benevolent sexism, but not receive more positive benefits.
Full-text available
The force factor method has garnered much attention and application in police use-of-force research, but the reliability of the method has yet to be intensively studied. Using official reports from the Seattle Police Department during a two-and-a-quarter-year period (n = 1,240), officer-suspect interactions were coded from the content of report narratives. Static force factors compared the maximum force applied by the officer with the maximum level of suspect resistance. Dynamic force factors were also recorded, including up to 10 iterations of dyadic action/reaction coded using the same coding scheme. The coding of force factors was completed independently by two teams working at different institutions in a fully crossed design. Evidence on the interrater reliability and subsequent utility of force factors is presented and discussed. Results indicate acceptable levels of agreement across coding teams and support the use of force factors as a central tool for studying asymmetrical social encounters and the proportionality of force.
Full-text available
Law enforcement agencies have a responsibility to control the use of force by police so as to minimally impact the liberty of citizens while ensuring both officer and public safety. Yet, there is an alarming disconnect between how the police, the public, and the courts view use of force. How can police administrators reliably distinguish justifiable applications of force from unnecessary force, or the use of excessive force, and do so in a manner that balances these differing views while ultimately adhering to legal doctrine? We propose a model and administrative process for identifying and addressing potential incidents of unnecessary force or use of excessive force by combining force factor methodology [Alpert, G. and Dunham, R. (1997). The Force Factor: Measuring Police Use of Force Relative to Suspect Resistance. Washington, DC: Police Executive Research Forum.] with the operationalization of the core test and mitigating factors delineated in Graham v. Conner.We provide a preliminary empirical test of this approach using administrative data on use of force incidents in the Seattle Police Department during a recent two-and-a-quarter year period. Results based upon both quantitative and qualitative analyses of administrative records suggest that the proposed model may assist in the identification of potentially excessive force incidents. Implications for policy and practice are discussed.
Although most police activities do not involve the use of force, those that do reflect important patterns of interaction between officer and citizen. After a brief survey of prior research, this study presents new data and findings to examine these patterns. The force factor applied and the sequential order of incidents of force is included in the analysis. The authors also examine police use of force from the suspect’s perspective, and create a new conceptual framework, the Authority Maintenance Theory, for examining and assessing police use of force.
This chapter discusses the role of anxiety in facilitating stereotypic judgment of outgroup behavior. Anxiety increases one's reliance on social stereotypes and makes the person traitor to his or her objective reason. When anxiety distracts persons from careful attention to the environment, they rely more on available cognitive structures such as social stereotypes in making judgments of others. Increased reliance on stereotypes manifests in two ways. First, stereotypes supply best guesses to complete gaps in information created by anxiety. Second, counter-stereotypic information loses impact when persons are anxious and their attention is restricted. In the research discussed in the chapter, stereotypes are beliefs about the characteristics or behaviors of most members of a social group.
Three studies tested basic assumptions derived from a theoretical model based on the dissociation of automatic and controlled processes involved in prejudice. Study 1 supported the model's assumption that high- and low-prejudice persons are equally knowledgeable of the cultural stereotype. The model suggests that the stereotype is automatically activated in the presence of a member (or some symbolic equivalent) of the stereotyped group and that low-prejudice responses require controlled inhibition of the automatically activated stereotype. Study 2, which examined the effects of automatic stereotype activation on the evaluation of ambiguous stereotype-relevant behaviors performed by a race-unspecified person, suggested that when subjects' ability to consciously monitor stereotype activation is precluded, both high- and low-prejudice subjects produce stereotype-congruent evaluations of ambiguous behaviors. Study 3 examined high- and low-prejudice subjects' responses in a consciously directed thought-listing task. Consistent with the model, only low-prejudice subjects inhibited the automatically activated stereotype-congruent thoughts and replaced them with thoughts reflecting equality and negations of the stereotype. The relation between stereotypes and prejudice and implications for prejudice reduction are discussed.
This chapter presents an integrated understanding of various impression formation processes. The chapter introduces a model of impression formation that integrates social cognition research on stereotyping with traditional research on person perception. According to this model, people form impressions of others through a variety of processes that lie on a continuum reflecting the extent to that the perceiver utilizes a target's particular attributes. The continuum implies that the distinctions among these processes are matters of degree, rather than discrete shifts. The chapter examines the evidence for the five main premises of the model, it is helpful to discuss some related models that raise issues for additional consideration. The chapter discusses the research that supports each of the five basic premises, competing models, and hypotheses for further research. The chapter concludes that one of the model's fundamental purposes is to integrate diverse perspectives on impression formation, as indicated by the opening quotation. It is also designed to generate predictions about basic impression formation processes and to help generate interventions that can reduce the impact of stereotypes on impression formation.