Content uploaded by Debbie S Ma
Author content
All content in this area was uploaded by Debbie S Ma on Apr 26, 2017
Content may be subject to copyright.
BASIC AND APPLIED SOCIAL PSYCHOLOGY, 35:515–524, 2013
Copyright © Taylor & Francis Group, LLC
ISSN: 0197-3533 print/1532-4834 online
DOI: 10.1080/01973533.2013.840630
When Fatigue Turns Deadly: The Association Between Fatigue
and Racial Bias in the Decision to Shoot
Debbie S. Ma
California State University, Northridge
Joshua Correll
University of Colorado at Boulder
Bernd Wittenbrink
University of Chicago
Yoav Bar-Anan
Ben-Gurion University of the Negev
N. Sriram and Brian A. Nosek
University of Virginia
Racial bias in the decision to shoot can be minimized if individuals have ample cognitive
resources to regulate automatic reactions. However, when individuals are fatigued,
cognitive control may be compromised, which can lead to greater racial bias in shoot/
don’t-shoot decisions. The current studies provide evidence for this hypothesis
experimentally using undergraduate participants (Study 1) and in a correlational design
testing police recruits (Study 2). These results shed light on the processes underlying the
decision to shoot and, given the high prevalence of fatigue among police officers, may
have important practical implications.
Correspondence should be sent to Debbie S. Ma, Department of
Psychology, California State University, 18111 Nordhoff Street,
Northridge, CA 91330. E-mail: debbie.ma@csun.edu
The shooting deaths of Amadou Diallo in 1999 and
Timothy Thomas in 2001—two unarmed Black men—by
police officers provoked intense public discussion and
prompted social psychological research investigating
potential causes. The role of suspect race in police offi-
cers’ decision to shoot was central to these discussions
and scientific inquiries (Correll, Park, Judd, &
Wittenbrink, 2002; Greenwald, Oakes, & Hoffman, 2003;
Payne, 2001; Plant, Peruche, & Butz, 2005). In addition,
researchers have identified a number of other factors that
influence the decision to shoot. For example, individual
differences in implicit associations linking Blacks to
weapons (Nosek et al., 2007; Payne, 2001) have been
shown to predict the decision to shoot (Correll, Park,
Judd, & Wittenbrink, 2007). Situational factors can also
be critical to understanding the decision to shoot.
Although the Diallo and Thomas cases were unrelated,
the circumstances surrounding them were similar in that
both shootings took place in threatening neighborhoods
and occurred late at night. External factors like neighbor-
hood and time of day are distinct from race but may still
have consequences for whether race figures into shoot/
don’t-shoot decisions. Research by Correll, Wittenbrink,
Park, Judd, and Goyle (2011), for example, suggests that
dangerous neighborhoods signal threat, which can lower
the threshold to shoot. The goal of the present article is
to investigate the role of another moderating factor—
fatigue—in the decision to shoot.
Downloaded by [University of Colorado at Boulder Libraries] at 10:25 26 December 2013
516 MA ET AL.
EXPERIMENTAL RESEARCH ON
THE ROLE OF SUSPECT RACE ON
THE DECISION TO SHOOT
Over the past decade, researchers have developed con-
trolled paradigms to isolate causal influences on the deci-
sion to shoot. For example, Correll and colleagues
developed a first-person shooter task (FPST; Correll et al.,
2002), in which participants are seated at a computer and
presented with a simplified video game where they assume
the role of a police officer surveilling public spaces. Images
such as train stations, apartment buildings, and parks are
displayed on the computer screen, and periodically a male
target appears. Targets are armed or unarmed Black or
White men. Participants are told to press one button to
indicate “shoot” when the target is armed and a different
button to indicate “don’t shoot” when the target is
unarmed. Race is peripheral to the task; however, the
FPST reveals racial bias in both error rates (e.g., partici-
pants mistakenly “shoot” unarmed targets more if they are
Black than White) and reaction times (e.g., participants are
faster to “shoot” armed targets if they are Black than
White). These effects are believed to emerge because of
cultural associations and stereotypes linking Blacks with
danger (Correll, Park, Judd, & Wittenbrink, 2007).
Research using the FPST and similar paradigms has
identified some of the cognitive processes that underlie
the decision to shoot. One of the critical findings to
emerge from this research involves the role of cognitive
control. Correll, Urland, and Ito (2006) measured event-
related brain potentials (ERPs) while participants com-
pleted the FPST. ERPs reflect electrical activity in the
brain (measured noninvasively by electrodes on the scalp)
in response to various stimuli. ERPs are useful because
they can reveal a lot about cognitive processing with high
temporal resolution. Correll and colleagues found that
individual differences in early ERP components that are
associated with cognitive control and response inhibition,
the N200, mediated the relationship between cultural ste-
reotypes linking Blacks to danger and racial bias in the
decision to shoot (Correll et al., 2006). Participants who
reported stronger cultural associations between Black
and danger showed smaller N200s in response to Black
targets compared to White targets, and this reduced
response inhibition resulted in greater bias on the FPST.
Complementary evidence comes from Payne’s (2001)
Weapons Identification Task (WIT). The WIT is similar
to the FPST but involves the rapid presentation of faces
(Black or White) followed by objects (guns or tools).
Participants’ task is to identify the objects as either guns
or tools with a key press. Like the FPST, the WIT typi-
cally reveals a robust pattern of racial bias, which is espe-
cially evident when participants are forced to respond
quickly (Payne, 2001). However, when the response
window is expanded, participants show significantly less
bias, presumably because they can recruit cognitive con-
trol processes necessary to override automatic associa-
tions. Research measuring ERPs while participants
completed the WIT emphasizes the importance of such
control in early processing (Amodio et al., 2004). Amodio
and colleagues examined an ERP component called the
event-related negativity (ERN), which originates from
the anterior cingulate cortex, a brain area that has been
linked to conflict detection. They found that the ERN
was larger on trials of the WIT where race and object
were stereotypically incongruent (e.g., a Black face fol-
lowed by a tool), suggesting that participants perceived a
conflict between race and object. Of interest, individuals
with more pronounced ERNs demonstrated greater accu-
racy and slower reaction times, perhaps because they
were exerting more effort to control their responses.
These studies converge on a central implication—people
may possess stereotypes associating Black men with weap-
ons or danger that could lead to racial bias in the decision
to shoot, but the availability of cognitive control may help
participants avoid responding in a stereotypical fashion.
For this reason, factors that diminish cognitive control,
such as fatigue, should increase racial bias in the decision
to shoot. Fatigue is characterized by both physical and
cognitive aspects and is an easy state to induce. Blagrove,
Alexander, and Horne (1995), for example, showed that
just 1 week of sleep reduction (i.e., sleeping 5 to even 7 h.
per night, as opposed to 8 hr) was sufficient to produce
fatigue (see also Banks & Dinges, 2007). Despite this,
people commonly underestimate the negative impact that
mild sleep loss can have on cognitive function (Banks &
Dinges, 2007). Physically fatigued individuals may experi-
ence a lack of energy, feelings of weakness, and sleepiness
(e.g., Shahid, Shen, & Shapiro, 2010). Mental symptoms
of fatigue include dullness and difficulty maintaining usual
levels of cognitive function (Chalder et al., 1993).
Those symptoms closely relate to cognitive depletion
and circadian rhythm (e.g., Lorist, Boksem, &
Ridderinkhof, 2005). Cognitive depletion is a temporary
state wherein one has diminished capacity to exert control
or volition over one’s affect, behavior, and cognition
(Muraven & Baumeister, 2000). Individuals who are
fatigued can experience cognitive depletion (Shahid et al.,
2010). Circadian rhythm is an individual difference and
characterizes regular fluctuations in circadian arousal
throughout the day. Periods of circadian arousal are asso-
ciated with greater processing capacity and working
memory efficiency, whereas lulls in circadian arousal are
associated with poorer cognitive ability (Folkard, Wever, &
Wildgruber, 1983) and feelings of fatigue (Shahid et al.,
2010). Because stereotypes serve as cognitive shortcuts that
allow individuals to quickly judge others (Macrae, Milne,
& Bodenhausen, 1994), individuals should rely on them to
a greater extent when they lack cognitive control.
Consistent with this logic, Govorun and Payne (2006)
Downloaded by [University of Colorado at Boulder Libraries] at 10:25 26 December 2013
FATIGUE AND THE DECISION TO SHOOT 517
demonstrated that depletion exacerbated racial bias in dis-
tinguishing weapons from tools using the WIT. Participants
performed either a long or relatively brief cognitively
taxing task prior to completing the WIT. Cognitively
depleted participants displayed significantly more racial
bias than control participants. Similarly, Bodenhausen
(1990) showed that stereotyping varied at different points
of the circadian rhythm. Participants judged the guilt of a
White or non-White (Hispanic or Black) defendant
accused of a transgression either at a resource-optimal or
suboptimal point of their circadian rhythm. Participants
judged non-White defendants more harshly than White
defendants, but only when participants were rendering
judgments during suboptimal periods.
The role of cognitive control and fatigue are especially
germane to police work and the typical circumstances of
officer-involved shootings. In 2000, the U.S. Department
of Justice assessed the prevalence of fatigue among police
officers (Vila, Kenney, Morrison, & Reuland, 2000).
Using a sample of 303 active officers from four different
police departments across the country, researchers found
that 41% of police officers were at clinical levels of sleep
deprivation. Using the FIT (“fitness-for-duty”) Workplace
Safety Screener (PMI, Inc.; Corfitsen, 1993), a physiologi-
cal assessment of involuntary saccadic velocity (i.e., the
speed with which the eye can track a moving point), Vila
and colleagues (2000) found that 19% of the officers
showed impairment. Alarmingly, 6.2% showed deficits
equivalent to the performance of a person with a .10%
blood alcohol concentration. The consequences of these
levels of fatigue among officers are important to consider
when it comes to cognitively demanding (and life-threat-
ening) tasks like determining whether a suspect is armed.
Taken together, the data suggest that officer fatigue
could harm decision making and increase racially biased
decisions. Indeed, sociological data support the hypothe-
sis that police officers’ fatigue is associated with shooting
behavior. In a national review of lethal force cases occur-
ring between 1976 and 1998, Geller (1982) asserted that
Blacks are disproportionately shot by officers and that
these shootings tend to occur during the evening hours
when officers are reportedly most tired (Vila & Kenney,
2002). Although this finding is consistent with the notion
that fatigue might lead officers to show racial bias in their
shooting, any number of factors could account for this
observation. For example, it may be more difficult to
make sense of threatening situations during the night
when Black suspects are involved. It is also possible that
Black criminals engage in more serious crimes at night
that warrant greater use of lethal force.
One way to address these alternatives is to conduct
laboratory-based experiments in which we can isolate
and manipulate fatigue and test for its influence on the
decision to shoot. Laboratory research investigating the
effects of fatigue on general decision making shows that
fatigue hampers the formation of good judgments, causes
people to persist with ineffectual task strategies, and
prompts impatient responses due to fatigue-related irrita-
bility (Staal, 2004; Vila et al., 2000). All of these processes
may contribute to the emergence of racial bias in the
decision to shoot with fatigue.
PRESENT RESEARCH
The current studies examined the effect of fatigue on the
decision to shoot. Study 1 used a sample of undergradu-
ate participants and tested whether cognitive depletion
influences racial bias on the FPST. Study 2 examined
police recruits and tested the relationship between perfor-
mance on the FPST and the amount of sleep each recruit
had the night before testing. Together, these studies con-
tribute to an important literature on the decision to shoot
and may further illuminate the cognitive processes impli-
cated in this decision. Moreover, although obviously sim-
ulated situations, these studies take steps toward
emulating the physiological conditions officers face in the
field to provide converging experimental evidence on a
real-world phenomenon for which direct experimental
testing is difficult.
STUDY 1
Method
Participants and design. Seventy-seven undergrad-
uate students (40 male, 36 female, one did not indicate
gender) at the University of Chicago participated in this
study in exchange for $10. Forty-four of the students
identified as White, 14 as Asian, 10 as Latino, five as
Black, three as other, and one did not indicate race.1 The
average age of the sample was 20.55 (SD = 2.66). The
study design was a 2 (depletion: depleted or control) × 2
(target race: Black or White) × 2 (object type: gun or
object) mixed model design with the last two factors vary-
ing within participant.
Procedure. Participants were randomly assigned to
either the depleted or control conditions and run individ-
ually. Following Govorun and Payne (2006), we used the
Stroop task (Stroop, 1935) to manipulate cognitive deple-
tion. The Stroop task is a response inhibition task in
which participants indicate the font color of words pre-
sented on the screen one at a time. Although this task is
easy when the text color and word are congruent (e.g.,
when the word RED is printed in red), the task becomes
1Analysis revealed no effects of participant gender or race on racial
bias in reaction time.
Downloaded by [University of Colorado at Boulder Libraries] at 10:25 26 December 2013
518 MA ET AL.
more difficult when the cues are incongruent (e.g., when
the word BLUE is printed in red; Friedman & Miyake,
2004). Because the task requires participants to control
responses to the irrelevant word cue and focus only on the
relevant color cue, the task is cognitively depleting.
Participants in the depleted condition completed 300
Stroop trials, whereas control participants completed 30
Stroop trials.
After participants finished the Stroop task, the experi-
menter administered the FPST. Each trial of the FPST
began with the presentation of a random number of back-
grounds (0–3), which appeared for a random duration
(500–800 ms). Next, a final background was randomly
selected and shown for a random duration (500–800 ms)
before a target appeared on the background. Participants
were instructed to respond to the target with a key press
on a keyboard. They were told that targets would be
armed (i.e., carrying a gun) or unarmed (i.e., carrying an
innocuous object like a wallet or cell phone). Guns were
always handguns and all objects (both guns and nonguns)
were either black or silver in color. If the target was armed,
participants were asked to press a button labeled “shoot.”
If the target was unarmed, participants were told to press
a button labeled “don’t shoot.” Consistent with previous
research (e.g., Correll, Park, Judd, & Wittenbrink, 2007),
we delivered feedback about accuracy on each trial along
with a running score. Participants earned 10 points for
correctly shooting armed targets and 5 points for indicat-
ing they would not shoot unarmed targets. Participants
were penalized 20 points for shooting an unarmed target
and 40 points for failing to shoot an armed target. The
score was given to motivate participants to respond accu-
rately and was not dependent on target race. From the
onset of the target, participants were given 850 ms to
respond. This time window is long enough that partici-
pants tend to show very high accuracy rates; therefore, the
key dependent variable was reaction time (Correll et al.,
2002). The FPST comprised 16 practice and 100 test trials.
Test trials featured 25 Black and 25 White unique male
targets. Each target appeared once armed and once
unarmed. Participants were thus presented with 25 Black
armed, 25 Black unarmed, 25 White armed, and 25 White
unarmed trials. Afterward, participants were debriefed
and thanked.
Results and Discussion
To analyze the reaction time data,2 we excluded trials on
which participants responded incorrectly (5.4%) or failed
2Data were also analyzed in terms of error rates; however, this analy-
sis yielded null results. Neither the control nor cognitively depleted par-
ticipants showed evidence for racial bias in terms of error rates,
sensitivity, or criterion. This is consistent with previous research that
has used the 850-ms version of the FPST and is likely attributable to the
longer response window, which minimizes variability in accuracy.
to respond within the 850 ms response window (2.6%).3
The remaining data were log-transformed and submitted
to a 2 (depletion: depleted or control) × 2 (target race:
Black or White) × 2 (object type: armed or unarmed)
mixed-model analysis of variance with repeated measures
on the last two factors. Although all analyses were con-
ducted using log-transformed data, means are presented
in ms to facilitate interpretation. Consistent with previous
research, we observed a significant main effect for Object
Type, F(1, 75) = 330.43, p < .001,
η
p
2 = .82. Participants
were faster to respond to armed relative to unarmed trials.
Neither the depletion, F(1, 75) = 3.09, p = .08,
η
p
2 = .04,
nor the target race main effects, F(1, 75) = 0.29, p = .59,
η
p
2 = .00, reached statistical significance. Although the
depletion main effect did not meet conventional levels of
statistical significance, depleted participants were margin-
ally faster than controls. The direction of the observed
trend was opposite of what we might expect, given
research showing that cognitive load slows reaction time
(Lamble, Kauranen, Laakso, & Summala, 1999).
However, we refrain from interpreting this effect given
that it might be an isolated effect and that the model con-
tains a significant higher order interaction (Crawford,
Jussim, & Pilanski, in press).
We also observed a significant Target Race × Object
Type interaction, F(1, 75) = 48.18, p < .001,
η
p
2 = .39. This
interaction reflects racial bias in the decision to shoot and
can be thought of in terms of faster responses to stereo-
type-congruent trials (Black armed and White unarmed)
than stereotype-incongruent trials (Black unarmed and
White armed). On unarmed trials, participants were faster
to respond if the target was White than Black, t(75) = 6.31,
p < .001,
η
p
2 = .34. Conversely, on armed trials, partici-
pants were faster to respond to Blacks than Whites,
t(75) = –4.14, p < .001,
η
p
2 = .19. The Depletion × Target
Race interaction was also significant, F(1, 75) = 5.85,
p = .02,
η
p
2 = .07. On Black target trials, depleted partici-
pants responded significantly faster than control partici-
pants, t(75) = –2.29, p = .02,
η
p
2 = .07. On White target
trials, there was no evidence that depleted and control
participants differed, t(75) = –1.09, p = .28,
η
p
2 = .02.
There was no evidence of a Depletion × Object Type inter-
action, F(1, 75) = 0.00, p = .98,
η
p
2 = .00.
The critical test of our hypothesis (that cognitive deple-
tion exacerbates racial bias in the decision to shoot) was
tested by the Depletion × Target Race × Object Type inter-
action. This interaction was statistically significant, F(1,
75) = 4.09, p = .05,
η
p
2 = .05 (see Figure 1). To better under-
stand the nature of this interaction, we examined the Target
Race × Object Type interaction separately for depleted and
for control participants. As previously described, the Target
3Participants in the depleted and control groups did not differ in
terms of the number of trials that were excluded due to incorrect
responses or time-outs.
Downloaded by [University of Colorado at Boulder Libraries] at 10:25 26 December 2013
FATIGUE AND THE DECISION TO SHOOT 519
Race × Object Type interaction reflects racial bias and can
be reduced to a single index, racial bias, using the following
equation: (RTBlack unarmed + RTWhite armed) – (RTBlack armed +
RTWhite unarmed) where higher values represent greater racial
bias. A similar analytic approach has been used in other
research using the FPST (Correll et al., 2002, Study 3).
Racial bias was significantly greater than zero among
control participants (Mracial bias = 19.38 ms, SD = 26.81),
t(36) = 4.24, p < .001,
η
p
2 = .33 and depleted participants
(Mracial bias = 33.78 ms, SD = 38.41), t(39) = 5.61, p < .001,
η
p
2 = .45. Although there was evidence of racial bias in
both conditions, the three-way interaction indicates that
racial bias was significantly more pronounced in the
depleted condition.
Next, we conducted simple effects tests to probe the
nature of the Target Race × Object Type interactions by
group. The simple effect of target race on unarmed trials
was significant for both control (Mdifference = 15.87 ms,
SD = 17.42) and depleted (Mdifference = 14.38 ms, SD = 23.77)
participants, t(36) = 5.52, p < .001,
η
p
2 = .46, and t(39) =
3.82, p < .001,
η
p
2 = .27, respectively. Participants in both
conditions were faster to accurately respond “don’t shoot”
when unarmed targets were White rather than Black. This
effect did not differ by depletion condition, t(75) = –0.20,
p = .84,
η
p
2 = .00. We then tested the simple effect of target
race on armed trials for each group separately. Although
this effect was not statistically significant among control
participants (Mdifference = 3.51 ms, SD = 24.51), t(36) = 0.98,
p = .33,
η
p
2 = .03, the effect was significant among depleted
participants (Mdifference = 19.40 ms, SD = 24.39), t(39) = 5.02,
p < .001,
η
p
2 = .39. Depleted participants were faster to
indicate “shoot” in response to armed targets if the target
was Black rather than White, t(75) = 2.80, p = .007,
η
p
2 = .10.
STUDY 2
Study 1 addressed the impact of one particular aspect of
fatigue, cognitive depletion. Although police work
is cognitively depleting, officer fatigue also stems from
systemic causes (Vila & Kenney, 2002) such as double
shifts (Bayley, 1994), overtime (Vila, 1996), disrupted
sleep patterns (Hockey, 1986; Mitler, Carskadon, Czeisler,
& Dement, 1988; Monk, 1990), required off-duty court
appearances (Kroes, 1985), shift irregularities (O’Neill &
Cushing, 1991; Pierce & Dunham, 1992), impaired recu-
peration (Gardell, 1987), and spillover of job-related
stress into personal life (Gardell, 1987). Fatigue due to
these factors may have psychological, emotional, and
physical consequences. In Study 2, we examined the con-
sequences of one aspect of fatigue—lack of sleep (Neylan
et al., 2010)—on shoot/don’t-shoot decisions. We hypoth-
esized that sleep would negatively relate to racial bias in
decisions to shoot.
Method
Participants and design. Participants were 224 new
recruits (174 male, 47 female, three unreported) to a large
metropolitan police department. The police department
volunteered to be part of a study examining the effects of
police academy training on the decision to shoot. Recruits
participated in the study on a voluntary basis. Eighty-
nine recruits were White, 73 were Latino/a, 27 were Black,
20 were Asian, five were biracial, one was multiracial,
and nine indicated other.4 Of the 213 recruits who
reported age, the average was 23.98 (SD = 2.96). The
study involved a 2 (target race: Black or White) × 2 (object
type: gun or object) × fatigue (continuously measured)
mixed-model design with the first two variables varying
within participant. Two measures of fatigue—amount of
sleep on the night prior to testing (sleep before testing)
and amount of sleep on an average night (average sleep)—
were assessed and were separately correlated with bias in
decisions to shoot.
Procedure. In the morning, during the first day of
training, recruits completed the FPST at a private website
using the Project Implicit infrastructure (http://implicit.
harvard.edu/). Compared to Study 1, we used a shorter
(700 ms) response window, which reduces variation in
reaction times and amplifies differences in error rates. We
used this version of the FPST, because at the time of test-
ing, we intended to track recruits as they progressed
through basic recruit training. Previous research sug-
gested that training (such as the type of training police
officers receive) tends to influence errors, rather than
reaction times (Correll, Park, Judd, Wittenbrink, Sadler,
et al., 2007). Thus we used the 700-ms version of the
FPST to collect a baseline measure of performance in
4Analyses by recruit gender and race revealed no differences in terms
of racial bias in reaction time, errors, d’, or c. The effects of sleep were
also not moderated by gender or race.
FIGURE 1 Means and standard errors for response reaction times as a
function of target race, object, and experimental condition (Study 1).
Downloaded by [University of Colorado at Boulder Libraries] at 10:25 26 December 2013
520 MA ET AL.
terms of errors. In addition, the change in the dependent
variable allowed us to test another measure of racial bias
in the decision to shoot, affording us an opportunity to
examine the generalizability of this relationship. The task
was otherwise identical to the paradigm in Study 1.
Two of the targets in the FPST (one unarmed White,
one unarmed Black) yielded extremely high error rates
(more than 70%) as compared to an average error rate of
31% for other targets. We therefore excluded both the
armed and unarmed trials featuring these targets from
the analyses. This left 96 test trials per participant. After
the FPST, recruits completed a questionnaire on which
they answered two questions about sleeping behavior:
“How many hours did you sleep last night?” (sleep before
testing) and “On average, how many hours do you sleep
each night?” (average sleep) Single-item measures of
hours slept on a typical night and number of hours slept
in the previous 24-hr period are commonly used to assess
fatigue (e.g., Veasey, Rosen, Barzansky, Rosen, & Owens,
2002). Rather than assess sleep from the previous 24 hr,
we opted to measure sleep the night before testing, as
testing occurred early in the morning. No other measures
of sleep or fatigue were assessed. On average, recruits
reported getting 6.65 h. (SD = 1.40) of sleep before test-
ing and 7.55 h. (SD = 0.99) average sleep.
Results and Discussion
Reaction time. Our first analysis focused on partici-
pants’ reaction times. We present these results for the sake
of completeness and caution that the 700-ms version of
FPST that was used is designed to maximize variability in
terms of the errors individuals make, rather than variabil-
ity in reaction time. For the reaction time analysis, we
treated data the same as in Study 1. We excluded trials on
which participants responded incorrectly (15.9%) or
failed to respond within the 700-ms response window
(22.6%). Note that we excluded approximately 10 times
the number of trials here as compared to Study 1. The
remaining data were log-transformed. Although the full
design of the study involved a 2 (target race: Black or
White) × 2 (object type: gun or object) × fatigue (continu-
ously measured and mean centered in primary analyses)
mixed-model with the first two variables varying within
participant, we simplified the design substantially by
reducing the Target Race × Object Type interaction to a
single measure of racial bias. Recall from Study 1 that
racial bias on the FPST represents a pattern of stereo-
type-congruent responding (e.g., faster shoot response to
armed Blacks and slower reaction time to armed Whites).
It is important to note that statistics resulting from the
full design are mathematically equivalent to the results of
this simpler model.
For the first analysis we regressed racial bias in reaction
time on sleep before testing. The intercept of this model,
which tests whether there was significant evidence of
racial bias, was statistically significant, F(1, 222) = 5.34,
p = .02,
η
p
2 = .02. However, there was no evidence of an
effect of sleep before testing on racial bias in reaction
times, F(1, 222) = 0.13, p = .72,
η
p
2 = .00. Overall, racial
bias in reaction times was observed in the sample of
recruits, but this was not moderated by sleep before test-
ing. In a second analysis, we regressed racial bias in reac-
tion times on average sleep. The test of the intercept was
identical, and there was no effect of average sleep on racial
bias in reaction time, F(1, 222) = 0.27, p = .60,
η
p
2 = .00.
Although these results do not replicate Study 1, we again
point out that this version of the FPST is designed to cap-
ture variance in errors and might not be well suited to
fairly test the hypothesis of bias in reaction time.
Errors. Our next set of analyses focused on the errors
that recruits made on the FPST. Again, the full design of
the study involved a 2 (target race: Black or White) × 2
(object type: gun or object) × fatigue (continuously mea-
sured) mixed-model with the first two factors varying
within participant. We again simplified the analysis by
computing racial bias, but this time in terms of errors
using the following equation: (ErrorBlack unarmed + ErrorWhite
armed) – (ErrorBlack armed + ErrorWhite unarmed). Racial bias in
errors can be conceptualized as making relatively few
mistakes on stereotype-congruent trials (armed Blacks
and unarmed Whites) and more mistakes on stereotype-
incongruent trials (unarmed Blacks and armed Whites).
First, we regressed racial bias in errors on sleep before
testing. The intercept of the model was statistically sig-
nificant, F(1, 222) = 11.70, p = .001,
η
p
2 = .05, meaning
that there was evidence of a racial bias in errors in the
sample (see Figure 2). We also observed a significant
effect of sleep before testing on racial bias, F(1, 222) = 3.77,
p = .05,
η
p
2 = .02 (see Figure 3). To better understand this
effect, we examined racial bias at low (−1 SD), average,
and high (+1 SD) levels of sleep before testing. For
recruits who reported low levels of sleep before testing,
we observed significant evidence of racial bias (Mracial
bias = .06, SD = .22), F(1, 222) = 14.14, p < .001,
η
p
2 = .06.
At mean levels of sleep before testing, racial bias was also
evident (Mracial bias = .03, SD = .15), F(1, 222) = 11.70,
p = .001,
η
p
2 = .05. However, at high levels of sleep before
testing, there was no evidence for racial bias (Mracial
bias = .01, SD = .21), F(1, 222) = 0.90, p = .34,
η
p
2 = .00. In
a second analysis, we examined the potential influence of
average sleep on racial bias in errors by regressing racial
bias on self-reported average sleep. We observed signifi-
cant racial bias in errors, as revealed by a significant
intercept (virtually identical to the previous analysis).
There was no evidence for an effect of average sleep on
racial bias, F(1, 222) = 0.81, p = .37,
η
p
2 = .00. Racial bias
was statistically equivalent across all levels of average
sleep.
Downloaded by [University of Colorado at Boulder Libraries] at 10:25 26 December 2013
FATIGUE AND THE DECISION TO SHOOT 521
Signal detection analysis. Next, we employed Signal
Detection Theory (SDT; Green & Swets, 1966) to model
recruits’ performance on the FPST. SDT is a statistical
technique that estimates two indices based on error rates.5
The first index derived from SDT is c, or the criterion,
which reflects the decision criterion. Above that criterion,
an individual will indicate a shoot response, but below
that criterion an individual will indicate a don’t shoot
response. Lower values of c therefore signify a more
5SDT d’ and c estimates are calculated using the following equa-
tions: d’ = zH – zFA; c = −0.5 × (zFA + zH). H represents the proportion
of hits relative to misses and FA represents the proportion of false
alarms relative to correct rejections. The z operator transforms these
terms to z-scores. H and FA are corrected to prevent infinite z scores.
When H or FA equal 0, a value of 1/2n is substituted, where n is the total
number of gun and object trials, respectively. When H or FA equal 1, a
value of 1 – (1/(2 × n)) is substituted.
“trigger happy” response. We examined whether fatigue
increases racial bias in the tendency to shoot, as reflected
by c by conducting a 2 Target Race (Black or White) × Sleep
Before Testing (continuously measured) regression with c
as the dependent variable. We observed a main effect of
target race, F(1, 222) = 13.44, p < .001,
η
p
2 = .06. On aver-
age, participants set a lower c for Black targets (Mc = –.09,
SD = .28) than White targets (Mc = –.01, SD = .28). This
main effect reflects racial bias. The main effect of sleep
before testing was not significant, F(1, 222) = 0.07, p = .79,
η
p
2 = .00. However, there was a Target Race × Sleep
Before Testing interaction, F(1, 222) = 4.01, p = .05,
η
p
2 =
.02 (see Figure 4).
To probe the nature of this interaction we tested the
difference between c for Black and White targets at low
(–1 SD), average, and high (+1 SD) levels of sleep before
testing. At low sleep before testing, there was a significant
effect of target race, t(222) = 3.97, p < .001,
η
p
2 = .07.
Participants set a lower c for Black targets (Mc = –.12,
SD = .40) than for White targets (Mc = .00, SD = .21). At
mean levels of sleep before testing, there was also a sig-
nificant target race effect, t(222) = 3.67, p < .001,
η
p
2 = .06.
Again, participants set a lower c for Black targets
(Mc = –.09, SD = .28) than for White targets (Mc = –.01,
SD = .28). Finally, at high levels of sleep before testing,
there was no evidence for an effect of target race,
t(222) = 1.07, p = .29,
η
p
2 = .01. Recruits set statistically
equivalent c for Black targets (Mc = –.07, SD = .40) and
White targets (Mc = –.03, SD = .40).
Our next analysis involved a 2 target race (Black or
White) × average sleep (continuously measured) regres-
sion with c as the dependent variable. We found a main
effect of target race, virtually identical to the prior analy-
sis. Again, participants set a lower threshold to shoot
Black (M = –.09, SD = .28) than White targets (M = –.01,
FIGURE 4 Means and standard errors for criteria (c) as a function of
target race, object, and self-reported number of hours slept the night
before testing (Study 2).
FIGURE 3 Means and standard errors for error rates as a function of
target race, object, and self-reported number of hours slept the night
before testing (Study 2).
FIGURE 2 Overall means and standard errors for error rate as a
function of target race and object (Study 2).
Downloaded by [University of Colorado at Boulder Libraries] at 10:25 26 December 2013
522 MA ET AL.
SD = .28). There was no evidence for an effect of average
sleep, F(1, 222) = 0.05, p = .83,
η
p
2 = .00 or a Target
Race × Average Sleep interaction, F(1, 222) = 0.47, p = .49,
η
p
2 = .00. These analyses indicate that the amount of sleep
recruits had prior to testing (but not on an average night)
had a significant influence on how much racial bias they
exhibited on the FPST. Those who reported sleeping less
the night prior to testing showed significant racial bias in
c, whereas racial bias was absent among those who
reported getting more sleep before testing.
The second index captured by SDT is d’, or sensitivity
in distinguishing between guns and nongun objects.
Higher d’ values indicate better performance. To examine
the relationship of fatigue on d’ we again conducted two
separate analyses looking first at sleep before testing and
then average sleep. A 2 (target race: Black or White) × Sleep
Before Testing (continuously measured) regression with d’
as the dependent variable revealed a main effect of target
race, F(1, 222) = 19.81, p < .001,
η
p
2 = .08, such that d’ was
higher for Black (M = 2.21, SD = .84) targets than White
targets (M = 2.02, SD = .83). This indicates better overall
performance to Black compared to White trials; however,
this effect is likely an anomaly. Race effects on d’ in the
FPST are typically null (Correll et al. 2002; Correll, Park,
Judd, Wittenbrink, Sadler, et al., 2007; Correll et al.,
2006), though they have also been observed in the oppo-
site direction than observed here (greater d’ for White tar-
gets than Black targets; Correll, Park, Judd, & Wittenbrink,
2007). There was no evidence for a main effect of Sleep
Before Testing, F(1, 222) = 0.10, p = .75,
η
p
2 = .00. The
Target Race × Sleep Before Testing interaction was also
not significant, F(1, 222) = 0.31, p = .58,
η
p
2 = .00.
Next, we conducted a 2 (target race: Black or
White) × Average Sleep (continuously measured) regres-
sion with d’ as the criterion. The main effect of target
race was essentially identical, such that participants
showed greater sensitivity to Blacks (M = 2.21, SD = .84)
than Whites (M = 2.02, SD = .83). There was no evidence
for an effect of average sleep, F(1, 222) = 0.01, p = .94,
η
p
2 = .00. The Target Race × Average Sleep interaction
was marginal, F(1, 222) = 3.50, p = .06,
η
p
2 = .02. For the
purposes of exploration, we decomposed this interaction
and examined the effect of target race at low (–1 SD),
average, and high (+1 SD) levels of average sleep. At low
levels of average sleep, there was a marginal target race
effect, t(222) = 1.81, p = .07,
η
p
2 = .02, such that partici-
pants showed greater d’ for Black (Md’ = 2.17, SD = 1.20)
than White targets (Md’ = 2.06, SD = 1.20). At mean
levels of average sleep, we found a significant effect of
target race, t(222) = 4.48, p < .001,
η
p
2 = .08. Sensitivity
toward Black targets (Md’ = 2.21, SD = 0.84) was greater
than to White targets (Md’ = 2.02, SD = 0.84). Finally, at
high levels of average sleep, d’ was again significantly
higher in response to Black targets (Md’ = 2.26, SD = 1.20)
compared to White targets (Md’ = 1.98, SD = 1.20),
t(222) = 4.48, p < .001,
η
p
2 = .08. The results revealed a
trending pattern, such that higher average sleep corre-
sponded with better discrimination of guns and nonguns
for Black compared to White targets. It may be that par-
ticipants who get more sleep on a typical night direct
more attention to the Black targets, allowing them to
identify objects more accurately on Black trials. Of
course, this is speculative and the current studies are not
designed to address this possibility. Furthermore, the
interaction was not statistically reliable so we caution
against overinterpreting this effect.
GENERAL DISCUSSION
Despite social psychologists’ collective knowledge of the
processes underlying the decision to shoot, additional
research examining moderating factors involved in this
high-stakes decision is needed. The current studies
explore the role that fatigue may have on the decision to
shoot. In Study 1 we experimentally manipulated cogni-
tive depletion and compared performance between con-
trol and cognitively depleted participants. Both groups
showed significant racial bias in response latencies, but
bias was even more pronounced among cognitively
depleted participants. Study 2 investigated the associa-
tion between sleep and decisions in the FPST among
police recruits. We found that recruits showed significant
racial bias in terms of reaction time, errors, criterion, and
sensitivity. Moreover, racial bias on errors and the crite-
rion was negatively associated with the amount of sleep
officers reported getting the night before testing and
racial bias. The effects reported in Study 2 are particu-
larly impressive in that we observed differences in shoot-
ing behavior within a relatively narrow range of sleep.
Recall that participants reported getting an average of
6.65 hours (SD = 1.40) of sleep before testing and 7.55 h.
(SD = 0.99) average sleep. That we observed significant
moderation in racial bias in errors and c by shifting about
1 h. in each direction of the mean is noteworthy.
The primary goal of the current research was to
examine the relationship between fatigue and the decision
to shoot. Given the prevalence of fatigue among police
officers (Vila et al., 2000), this is an important question
with real world implications. Although the current
research is suggestive of the fact that fatigue exacerbates
racial bias in the decision to shoot, we point out limitations
of the current research. First and foremost, the
correlational design of Study 2 leaves the results vulnerable
to alternative interpretations. For instance, officers who
received less sleep the night before testing might have slept
less because they were more anxious and/or aroused.
After all, testing for Study 2 took place on the 1st day of
basic recruit training, which could very likely have induced
arousal among some participants. Arousal has been shown
to increase stereotyping and prejudice (Kim & Baron,
1988; Lambert et al. 2003; Wilder, 1993). Participants who
Downloaded by [University of Colorado at Boulder Libraries] at 10:25 26 December 2013
FATIGUE AND THE DECISION TO SHOOT 523
slept less the night before testing may also differ in terms
of important personality traits that bear on stereotyping.
People who score higher on neuroticism, for example, also
tend to sleep less (Kumar & Vaidya, 1982) and may also
have a greater tolerance for stereotyping (Carter, Hall,
Carney, & Rosip, 2006). Ultimately, although we have
attempted to measure fatigue in Study 2, we cannot fully
rule out alternative interpretations with respect to this
study. However, these concerns are partially allayed by the
fact that we found a causal relationship between fatigue
and bias in the decision to shoot in Study 1, albeit with a
distinct form of fatigue—cognitive depletion rather than
amount of sleep. The basic assumption motivating the
current research is the notion that the decision to shoot is
sensitive to factors that vary across time and situation. As
such, it is important to consider that the effects of fatigue
may also be moderated by other factors. For example, the
extensive training and police officers receive may weaken
the effect of fatigue. A recent meta-analysis by Hagger,
Wood, Stiff, and Chatzisarantis (2010) found that
individuals who received self-control training were less
affected by cognitive depletion than those who did not
receive self-control training. One reason for why this
might be is that training allows individuals to automate
responding to relevant information (i.e., the object targets
are holding) while tuning out irrelevant information (i.e.,
race). Consistent with this idea, Plant et al. (2005)
proposed that training can eliminate bias. However, the
attenuation of racial bias in the decision to shoot may
depend largely on the type of outcomes being measured.
For example, research conducted on police officers finds
that although officers show a robust pattern of racial bias
on reaction times they do not show any evidence of racial
bias in the mistakes they make (Correll, Park, Judd,
Wittenbrink, Sadler, et al., 2007). In other words, although
police officers are slower to respond to stereotype
incongruent trials (i.e., armed Whites and unarmed
Blacks) than stereotype congruent trials (i.e., unarmed
Whites and armed Blacks), they end up making decisions
that are not influenced by target race. This strongly
suggests that police officers are still activating cultural
stereotypes associating Blacks with danger but somehow
manage to override these stereotypes and respond to the
object targets are holding.
Given that officers rely heavily on cognitive control to
override racially biased responding, training may produce
an ironic effect whereby trained individuals are actually
more racially biased when fatigued or when their execu-
tive control is disrupted in other ways. We recently
observed evidence of this in a study we conducted exam-
ining the effect of cognitive load on training in the FPST
(Correll, Wittenbrink, Axt, Goyle, & Miyake, 2013). Like
fatigue, cognitive load can disrupt cognitive control (Lavie,
Hirst, de Fockert, & Viding, 2004). Participants received
training on the FPST (experts) or not (novices) and then
all participants completed the FPST under three cognitive
load conditions: no load, low load, and high load. Load
was manipulated using a concurrent task that varied in
difficulty. Replicating previous research (Correll, Park,
Judd, Wittenbrink, Sadler, et al., 2007), when there was no
load, experts showed significantly less racial bias than
novices in terms of c. Under low cognitive load, experts
and novices showed the same degree of racial bias in c. Of
interest, cognitive load led to increased bias for experts
but had essentially no effect on novices. This is consistent
with the idea that training may make individuals more
susceptible to racial bias when their cognitive resources
are compromised.
In the field, fatigue, training, and many other variables
(e.g., arousal, neighborhood, the type of call to which
officers are responding, etc.) presumably interact to influ-
ence a police officer as he or she decides whether to use
lethal force. Identifying these contributing factors and
empirically investigating how they interact to inform this
decision is obviously important. Although isolating and
manipulating variables in a controlled, laboratory setting
is necessary for establishing causality, determining
whether these relationships hold up in the field requires
studying these relationships in the real world (e.g., analy-
sis of sociological data). Although this is true of most
social psychological research, it may be essential for
studying the decision to shoot where realistically recreat-
ing the decision to shoot context would be virtually
impossible not to mention unethical. Thus, bridging the
gap between the lab and the field is a necessary step in
understanding the complexities involved in police offi-
cers’ decisions to shoot.
ACKNOWLEDGMENTS
Primary support for this work was provided by National
Science Foundation Continuing Grant 0642580 to the
second and third authors. Additional support was pro-
vided by the Booth School of Business. We thank Miho
Goto, Elton Lor, and Ashley Meyer for their invaluable
help with this research.
REFERENCES
Amodio, D. M., Harmon-Jones, E., Devine, P. G., Curtin, J. J., Hartley,
S. L., & Covert, A. E. (2004). Neural signals for the detection of race
bias. Psychological Science, 15, 88–93.
Banks, S., & Dinges, S. F. (2007). Behavior and physiological conse-
quences of sleep restriction. Journal of Clinical Sleep Medicine, 3,
519–528.
Bayley, D. (1994). Police for the future. New York: NY: Oxford.
Blagrove, M., Alexander, C., & Horne, J. A. (1995). The effects of
chronic sleep reduction on the performance of cognitive tasks sen-
sitive to sleep deprivation. Applied Cognitive Psychology, 9, 21–40.
Bodenhausen, G. V. (1990). Stereotypes as judgmental heuristics:
Evidence of circadian variations in discrimination. Psychological
Science, 1, 319–322.
Downloaded by [University of Colorado at Boulder Libraries] at 10:25 26 December 2013
524 MA ET AL.
Carter, J. D., Hall, J. A., Carney, D. R., & Rosip, J. C. (2006). Individual
differences in the acceptance of stereotyping, Journal of Research in
Personality, 40, 1103–1118.
Chalder, T., Berelowitz, G., Pawlikowska, T., Watts, L., Wessely, S.,
Wright, D., & Wallace, E. P. (1993). Development of a fatigue scale.
Journal of Psychosomatic Medicine, 37, 147–153.
Corfitsen, M. (1993). Tiredness and visual reaction time among night-
time cab drivers: A roadside survey. Accident Analysis and Prevention,
25, 667–673.
Correll, J., Park, B., Judd, C. M., & Wittenbrink, B. (2002). The police
officer’s dilemma: Using ethnicity to disambiguate potentially threat-
ening individuals. Journal of Personality & Social Psychology, 83,
1314–1329.
Correll, J., Park, B., Judd, C. M., & Wittenbrink, B. (2007). The influ-
ence of stereotypes on decisions to shoot. European Journal of Social
Psychology, 37, 1102−1117.
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 & Social
Psychology, 92, 1006−1023.
Correll, J., Urland, G. L., & Ito, T. A. (2006). Event-related potentials
and the decision to shoot: The role of threat perception and cogni-
tive control. Journal of Experimental Social Psychology, 42, 120−128.
Correll, J., Wittenbrink, B., Axt, J., Goyle, A., & Miyake, A. (2013).
Executive officers? Cognitive control over racial bias in the deci-
sion to shoot. Unpublished manuscript, University of Chicago,
Chicago, IL.
Correll, J., Wittenbrink, B., Park, B., Judd, C. M., & Goyle, A. (2011).
Dangerous enough: Moderating racial bias with contextual threat
cues. Journal of Experimental Social Psychology, 47, 184–189.
Crawford, J. T., Jussim, L., & Pilanski, J. M. (in press). How (not) to
interpret and report main effects and interactions in multiple regres-
sion: Why Crawford and Pilanski (2013) did not actually replicate
Lindner and Nosek (2009). Political Psychology.
Folkard, S., Wever, R. A., & Wildgruber, C. M. (1983). Multioscillatory
control of circadian rhythms in human performance. Nature, 305,
223–226.
Friedman, N. P., & Miyake, A. (2004). The relations among inhibition
and interference control functions: A latent-variable analysis. Journal
of Experimental Psychology: General, 133, 101–139.
Gardell, B. (1987). Efficiency and health hazards in mechanized work.
In J. Quick, R. Bhagat, J. Dalton, & J. Quick (Eds.), Work stress:
Health care systems in the workplace. New York, NY: Praeger.
Geller, W. A. (1982). Deadly force: What we know. Journal of Police
Science and Administration, 10, 151−177.
Govorun, O., & Payne, K. B. (2006). Ego-depletion and prejudice: Sepa-
rating automatic and controlled components. Social Cognition, 24,
111–136.
Green, D. M., & Swets, J. A. (1966). Signal detection theory and psycho-
physics. New York, NY: Wiley. (Reprinted 1974, Huntington, NY:
Krieger)
Greenwald, A. G., Oakes, M. A., & Hoffman, H. G. (2003). Targets
of discrimination: Effects of race on responses to weapons holders.
Journal of Experimental Psychology, 39, 399−405.
Hagger, M. S., Wood, C., Stiff, C., & Chatzisarantis, N. L. D. (2010).
Ego depletion and the strength model of self-control: A meta-analy-
sis. Psychological Bulletin, 136, 495–525.
Hockey, G. (1986). Changes in operator efficiency as a function of envi-
ronmental stress, fatigue, and circadian rhythms. In K. R. Boff, L.
Kaufman, & J. Thomas (Eds.), Handbook of perception and human
performance, Vol. II: Cognitive process and performance (pp. 1–49).
New York, NY: John Wiley.
Kim, H. S., & Baron, R. S. (1988). Exercise and the illusory correlation:
Does arousal heighten stereotypic processing? Journal of Experimen-
tal Social Psychology, 24, 366–380.
Kroes, W. (1985). Society’s victims—The police: An analysis of job stress
in policing, (2nd ed.). Springfield, IL: Thomas.
Kumar, A., & Vaidya, A. K. (1982). Neuroticism in short and long
sleepers. Perceptual and Motor Skills, 54, 962–962.
Lambert, A. J., Payne, B. K., Jacoby, L. L., Shaffer, L. M., Chasteen,
A. L., & Khan, S. R. (2003). Stereotypes as dominant responses: On
the “social facilitation” of prejudice in anticipated public contexts.
Journal of Personality and Social Psychology, 84, 277–295.
Lamble, D., Kauranen, T., Laakso, M., & Summala, H. (1999).
Cognitive load and detection thresholds in car following situations:
Safety implication for using mobile (cellular) telephones while
driving. Accident Analysis and Prevention, 31, 617–623.
Lavie, N., Hirst, A., de Fockert, J. W., & Viding, E. (2004). Load theory
of selective attention and cognitive control. Journal of Experimental
Psychology: General, 133, 339–354.
Lorist, M., Boksem, M., & Ridderinkhof, K. (2005). Impaired cognitive
control and reduced cingulate activity during mental fatigue. Cogni-
tive Brain Research, 24, 199–205.
Macrae, C. N., Milne, A. B., & Bodenhausen, G. V. (1994). Stereotypes
as energy-saving devices: A peek inside the cognitive toolbox. Journal
of Personality and Social Psychology, 66, 37–47.
Mitler, M., Carskadon, M., Czeisler, C., & Dement, W. (1988). Catastro-
phes, sleep, and public policy: Consensus report. Sleep, 11, 100–109.
Monk, T. (1990). Shiftworker performance. Occupational Medicine:
State of the Art Reviews, 5, 183–198.
Muraven, M., & Baumeister, R. F. (2000). Self-regulation and depletion
of limited resources: Does self-control resemble a muscle? Psycho-
logical Bulletin, 126, 247–259.
Neylan, T. C., Metzler, T. J., Henn-Haase, C., Blank, Y., Tarasovsky, G.,
McCaslin, S. E., et al. (2010). Prior night sleep duration is associated
with psychomotor vigilance in a healthy sample of police academy
recruits. Chronobiology International, 27, 1493–1508.
Nosek, B. A., Smyth, F. L., Hansen, J. J., Devos, T., Lindner, N. M.,
Ratliff (Ranganath), K. A., … Banaji, M. R. (2007). Pervasiveness
and correlates of implicit attitudes and stereotypes. European Review
of Social Psychology, 18, 36–88.
O’Neill, J., & Cushing, M. (1991). The impact of shift work on police
officers. Washington, DC: Police Executive Research Forum.
Payne, B. K. (2001). Prejudice and perception: The role of automatic
and controlled processes in misperceiving a weapon. Journal of
Personality & Social Psychology, 81, 181−192.
Pierce, J. & Dunham, R. (1992). The 12-hour work day: A 48-hour,
eight-day week. Academy of Management Journal, 35, 1086–1098.
Plant, E. A., Peruche, B. M., & Butz, D. A. (2005). Eliminating auto-
matic racial bias: Making race non-diagnostic for responses to crimi-
nal suspects. Journal of Experimental Social Psychology, 41, 141−156.
Shahid, A., Shen, J., & Sharpiro, C. M. (2010). Measurements of sleepi-
ness and fatigue. Journal of Psychosomatic Research, 69, 81–89.
Staal, M. A. (2004). Stress and suicide in the military: Service-specific
risk factors. In S. P. Shohov (Ed.), Advances in psychology research,
Vol. 30 (pp. 155–192). Hauppauge, NY: Nova Science.
Stroop, J. R. (1935). Studies of interference in serial verbal reactions.
Journal of Experimental Psychology, 18, 643–662.
Veasey, S., Rosen, R., Barzansky, B., Rosen, I., & Owens, J. (2002). Sleep
loss and fatigue in residency training. Journal of the American Medi-
cal Association, 288, 1116–1124.
Vila, B., (1996). Tired cops: Probable connections between fatigue and
the performance, health, and safety of patrol officers. American
Journal of Police, 2, 51–92.
Vila, B., & Kenney, D. J. (2002). Tired cops: The prevalence and potential
consequences of fatigue. National Institute of Justice Journal, 248, 17–31.
Vila, B. J., Kenney, D. J., Morrison, G. B., & Reuland, M. (2000). Evalu-
ating the effects of fatigue on police patrol officers: Final report. Wash-
ington, DC: U.S. Department of Justice, National Criminal Justice
Reference Service.
Wilder, D. A. (1993). The role of anxiety in facilitating stereotypic judg-
ments of outgroup behavior. In D. M. Mackie & D. L. Hamilton
(Eds.), Affect, cognition, and stereotyping: Interactive processes in
group perception (pp. 87–109). San Diego, CA: Academic.
Downloaded by [University of Colorado at Boulder Libraries] at 10:25 26 December 2013
