Content uploaded by Natalie M. Daumeyer
Author content
All content in this area was uploaded by Natalie M. Daumeyer on Sep 15, 2020
Content may be subject to copyright.
Contents lists available at ScienceDirect
Journal of Experimental Social Psychology
journal homepage: www.elsevier.com/locate/jesp
Consequences of attributing discrimination to implicit vs. explicit bias
☆
Natalie M. Daumeyer
⁎
, Ivuoma N. Onyeador, Xanni Brown, Jennifer A. Richeson
Yale University, Department of Psychology, United States of America
ARTICLE INFO
Keywords:
Implicit bias
Bias attribution
Accountability
Science communication
ABSTRACT
Implicit bias has garnered considerable public attention, with a number of behaviors (e.g., police shootings)
attributed to it. Here, we present the results of 4 studies and an internal meta-analysis that examine how people
reason about discrimination based on whether it was attributed to the implicit or explicit attitudes of the per-
petrators. Participants' perceptions of perpetrator accountability, support for punishment, level of concern about
the bias, and support for various efforts to reduce it (e.g., education) were assessed. Taken together, the results
suggest that perpetrators of discrimination are held less accountable and often seen as less worthy of punishment
when their behavior is attributed to implicit rather than to explicit bias. Moreover, at least under some cir-
cumstances, people express less concern about, and are less likely to support efforts to combat, implicit com-
pared with explicit bias. Implications for efforts to communicate the science of implicit bias without under-
mining accountability for the discrimination it engenders are discussed.
Since 2015, implicit bias has been mentioned in over 13,000 news
articles, currently yields over 40 million Google search results, and was
even featured in a recent episode of the popular television show Grey’s
Anatomy (Clack, Rhimes, & Sullivan, 2018). Despite this public atten-
tion, little is known about how people incorporate information about
implicit bias into their reasoning about discrimination. While many
researchers, activists, and reporters presume that educating the public
about implicit bias will galvanize support to combat its discriminatory
consequences, there is reason to expect greater awareness of implicit
bias may reduce the extent to which people hold others accountable for
the discrimination it engenders (Cameron, Payne, & Knobe, 2010). The
present research considers this question.
1. Implicit vs. explicit bias and moral reasoning
Implicit biases are associations and reactions that emerge auto-
matically and often without awareness upon encountering a relevant
stimulus (Gawronski & Bodenhausen, 2006;Greenwald & Banaji,
1995). In the social domain, for instance, stereotypical concepts such as
“criminal”and “dangerous”automatically come to mind for many
people when they encounter or are exposed to images of young black
men (Eberhardt, Goff, Purdie, & Davies, 2004;Hester & Gray, 2018).
And, people typically unconsciously hold more negative attitudes or
feelings about racial/ethnic outgroup, compared with ingroup, mem-
bers (Axt, Ebersole, & Nosek, 2014). These implicit forms of bias stand
in contrast to more explicit forms (Carter & Murphy, 2015;Sommers &
Norton, 2006), including preferences, beliefs, and attitudes of which
people are generally consciously aware and can, when willing, identify
and communicate to others (Dovidio & Gaertner, 2010). Importantly,
research suggests that both implicit and explicit bias can shape the
judgments and decisions we make (e.g., Bertrand & Mullainathan,
2004;Devine, 1989), as well as how we behave, albeit often in different
ways (Dasgupta, 2004;Dovidio, Kawakami, & Gaertner, 2002). But,
whereas people are generally aware of the influence their explicit at-
titudes have on their behavior, they are often unaware of the influence
their implicit biases can have (Dovidio et al., 2002).
Conscious awareness, in other words, is a key element that distin-
guishes implicit from explicit bias. In turn, it is possible that people are
likely to hold others less accountable for discriminatory behavior that is
thought to be due to implicit, rather than explicit, attitudes. Theoretical
and empirical work on moral reasoning argues that in order to be
morally responsible for an action, the actor needs to have some level of
awareness of and control over their behavior (e.g., Alicke, 2000;Nadler
& McDonnell, 2012;Shaver, 1985). Similarly, both classic and more
recent models of behavioral attribution (Heider, 1958;Malle, 1999;
Malle, Guglielmo, & Monroe, 2014;Monroe & Malle, 2017;Weiner,
1995) assert the critical role of perceived intent in shaping these jud-
gements. In other words, the mental state of the actor is essential to an
assessment of culpability (Cameron et al., 2010;Cushman, 2015).
Consequently, we generally judge perpetrators of harmful actions more
https://doi.org/10.1016/j.jesp.2019.04.010
Received 2 August 2018; Received in revised form 28 March 2019; Accepted 29 April 2019
☆
This paper has been recommended for acceptance by Sarah J Gervais.
⁎
Corresponding author.
E-mail address: natalie.daumeyer@yale.edu (N.M. Daumeyer).
Journal of Experimental Social Psychology 84 (2019) 103812
0022-1031/ © 2019 Published by Elsevier Inc.
T
negatively when they are thought to have engaged in those acts
knowingly, consciously, and/or intentionally (see Heider, 1958;Knobe
& Nichols, 2011).
How might implicit bias attributions for discrimination shape
judgments of accountability because of inferences about the mental
state of the perpetrator? To the extent that implicit bias is understood as
being largely unconscious (i.e., outside of awareness) and/or gives rise
to behavior that is unintentional, these theories of attribution suggest
that people who engage in discrimination that is due to implicit bias
should be held less culpable for their actions than those who engage in
discrimination due to explicit bias. Consistent with this prediction, re-
search has found an actor's awareness of their bias to shape the extent
to which perceivers hold them culpable for discrimination born of that
bias (Cameron et al., 2010;Redford & Ratliff, 2016). When presented
with scenarios wherein a manager is said to discriminate against Black
people when making hiring or promotion decisions, for instance, per-
ceivers found the manager less morally responsible (blameworthy, ac-
countable) if he was said to be unaware, rather than aware, of his ne-
gative attitudes toward Black people (Cameron et al., 2010;Redford &
Ratliff, 2016).
Building on this work, the present research considers how people
respond to communications of scientificfindings that reveal the dis-
criminatory effects of implicit (rather than explicit) bias. Similar to
these recent studies, as well as how researchers often discuss implicit
bias in the literature (Casad, Flores, & Didway, 2013;Greenwald &
Banaji, 1995) and to the public (Payne, Niemi, & Doris, 2017), we focus
on awareness (or the lack thereof) as the key factor that distinguishes
implicit from explicit forms of bias. Rather than asking participants to
respond to a scenario describing discrimination by a single perpetrator
(see, again, Cameron et al., 2010;Redford & Ratliff, 2016), however,
we present participants with an ostensible news article that details the
findings of recent research wherein patterns of discriminatory behavior
were observed by categories of perpetrators (e.g., doctors, police offi-
cers) and attributed either to their implicit or explicit attitudes. To the
extent that implicit bias is understood to involve a lack of awareness of
one's attitudes (and/or its influence on behavior), even media reports of
scientific studies revealing systemic discrimination should result in the
perpetrators being held less accountable if their acts are attributed to
implicit, rather than explicit, bias.
The present research also sought to extend the relatively nascent
body of research on the consequences of implicit, compared with ex-
plicit, bias attribution in two additional ways. First, instead of the overt
cases of racial discrimination (e.g., failure to hire/promote Black
people) primarily considered in past work, the present research ex-
amined responses to subtle differential behavioral treatment in the
context of professional (i.e., medical, law enforcement) interactions.
Second, in addition to examining the perceived responsibility of per-
petrators (i.e., accountability, support for punishment), we also ex-
plored whether perceivers report differential concern about implicit
compared to explicit bias, support reform efforts to mitigate the effects
of implicit compared to explicit bias on behavior, and hold relevant
institutional actors (e.g., police departments) differentially accountable
for discrimination that is attributed to implicit, relative to explicit, bias.
Perhaps, while implicit bias may reduce individual culpability for dis-
criminatory behavior, it may not affect the perceived responsibility of
institutional actors to combat discrimination that occurs in their in-
stitutions.
2. The present work
Four studies and an internal meta-analysis consider how people
reason about communication of the science of implicit (vs. explicit) bias
across a variety of social identity dimensions (political affiliation, age,
race) and settings (interactions between doctors and patients, police
officers and citizens). Consistent with past work (Cameron et al., 2010;
Redford & Ratliff, 2016), we predicted that perpetrators of
discrimination would be held less accountable and less worthy of
punishment when their discrimination is attributed to their implicit
rather than explicit attitudes. We also considered whether this bias
attribution shapes perceivers' level of concern about discrimination and
the extent to which they support reform efforts to combat it.
3. Studies 1A and 1B
Studies 1A and 1B examined the effects of implicit vs. explicit bias
attribution in the context of discrimination by medical doctors toward
their patients based on political attitudes. For each study, data collec-
tion was completed prior to analyzing the data.
3.1. Methods
All materials and data for all studies are available at OSF (osf.io/
5gtyv). All measures, manipulations, and exclusions in the studies are
reported in the manuscript.
3.1.1. Participants
Through TurkPrime (Litman, Robinson, & Abberbock, 2017), we
recruited Amazon Mechanical Turk workers in the United States to
complete each study in exchange for $1.00. In Study 1A, 273 were
recruited, however two participants were excluded for admitting to
answering randomly and four were excluded due to suspicion, resulting
in a sample of 267 participants (43.8% female, 71.5% White,
M
age
= 34.92; 151 in the explicit condition). In Study 1B, 300 partici-
pants were recruited, but one was excluded for admitting to answering
randomly, leaving in a sample of 299 participants (47.5% female,
74.2% White, M
age
= 35.00; 149 in the explicit condition). The sample
size in Study 1A was chosen based on an a priori decision to enroll a
sufficient number of participants to yield an analysis sample with at
least 100 people per condition. This target number was increased to 150
for Study 1B, given that it was a replication attempt.
3.1.2. Bias manipulation
Participants read an article inspired by recent research about doc-
tors who demonstrate bias toward their patients based on political
ideology (Hersh & Goldenberg, 2016). The article described a study in
which both Democratic and Republican doctors exhibited bias toward
patients who engaged in somewhat politicized health behaviors—gun
ownership and recreational marijuana use. In both conditions, the be-
havioral consequences of the bias were identical: doctors spent less time
with and exhibited more aggressive body language toward patients they
had biases against (i.e., marijuana users or gun owners). Similar to past
work, the awareness dimension of explicit and implicit bias was ma-
nipulated (Cameron et al., 2010;Redford & Ratliff, 2016). In the ex-
plicit bias condition, participants read that “doctors were somewhat
aware they were treating patients differently, but thought they needed
to be tough with their patients in order to encourage them to re-eval-
uate their behavioral choices and ultimately live a healthier lifestyle.”
In the implicit condition, implicit bias was first defined as “attitudes or
stereotypes that affect our understanding, actions, and decisions in
ways that we are typically not aware of.”Participants also read that
“the doctors had no conscious knowledge that they were treating pa-
tients any differently based on their political views.”
3.1.3. Dependent variables
All items were rated on Likert scales from 1 (strongly disagree) to 7
(strongly agree).
3.1.3.1. Accountability. Accountability for bias was measured using six
items (α
1A
= 0.83; α
1B
= 0.82; e.g., “Doctors should be held
responsible for any biases they have that may impact how they
interact with patients”). Higher scores indicate greater accountability.
1
N.M. Daumeyer, et al. Journal of Experimental Social Psychology 84 (2019) 103812
2
3.1.3.2. Punishment. Punishment for bias was measured using three
items (α
1A
= 0.86; α
1B
= 0.89; e.g., “Doctors who repeatedly
demonstrate biases toward patients should have their license to
practice medicine suspended”). Higher scores indicate greater support
for punishment.
3.1.3.3. Concern. Concern about the bias was measured using six items
(α
1A
= 0.84; α
1B
= 0.86; e.g., “The bias I read about in the article is
concerning”). Higher scores indicate greater concern.
3.1.3.4. Reform. Support for reform was measured using five items
(α
1A
= 0.82; α
1B
= 0.83; e.g., “Doctors should be required to undergo
training to prevent their biases from impacting their treatment of
patients”). Higher scores indicate greater support for reform.
3.1.4. Potential moderator variables
We administered several potential moderator variables (e.g., bias
awareness, Perry, Murphy, & Dovidio, 2015). Across all studies, none of
these scales consistently moderated the effects reported below, and
thus, they are not discussed further. Again, for all materials and data see
osf.io/5gtyv.
3.2. Procedure
Participants provided informed consent, and then were randomly
assigned to the explicit or implicit bias attribution condition. After
reading the article, they reported their perceptions of accountability,
concern about the bias, support for reform, and support for punishment,
in this order. Items within scales were randomized across participants.
Participants next completed a number of potential moderator variables,
then reported their demographic information (e.g., race, age, gender,
political ideology, and whether they own a gun or engage in recrea-
tional marijuana use). Participants also completed data quality and
attention checks, and then were thanked, debriefed, and paid.
3.3. Results
We conducted an independent samples t-test exploring the effect of
the bias attribution manipulation (explicit vs. implicit) on each de-
pendent measure.
2
For each outcome, we present the results for Study
1A followed by those for Study 1B (the replication sample). For Study
1A, the analyses had 80% power to detect an effect size of Cohen's
d= 0.35. For Study 1B, the analyses had 80% power to detect an effect
size of d= 0.33. Correlations among the dependent variables for each
study are provided in Table 1. The means and standard deviations for
each dependent variable by condition are provided in Table 2.
3.3.1. Accountability
Analyses revealed that there was unequal variance for perceptions
of accountability among the two conditions in Study 1A, F
(1,265) = 6.46, p= .012. Thus, we provide t- and p-values for unequal
variances assumed. Consistent with predictions, there was a significant
difference in accountability as a function of bias attribution in Study
1A, t(221.39) = 3.51, p= .001, d= 0.44. Participants in the implicit
bias condition held doctors less accountable than participants in the
explicit bias condition. This effect replicated in Study 1B, t
(297) = 5.00, p< .001, d= 0.58.
3.3.2. Punishment
Contrary to predictions, in Study 1A, there was no difference in
support for punishment as a function of bias attribution, t(265) = 1.01,
p= .312, d= 0.13. In the replication sample (Study 1B), however,
support for punishment did differ by condition, t(297) = 3.91,
p< .001, d= 0.45. Participants in the implicit bias condition were
significantly less supportive of punishment than participants in the
explicit bias condition.
3.3.3. Concern
In Study 1A, there was no difference in participants' concern about
bias as a function of condition, t(265) = 1.08, p= .281, d= 0.13. In
Study 1B, however, concern about bias did differ significantly by con-
dition, t(297) = 3.42, p= .001, d= 0.40, such that participants in the
implicit bias condition expressed less concern than participants in the
explicit bias condition.
3.3.4. Reform
In Study 1A, there was no difference in support for reform efforts as
a function of bias attribution, t(265) = 0.80, p= .427, d= 0.10. The
effect in Study 1B was also not reliable, t(297) = 1.43, p= .155,
d= 0.16.
3.4. Discussion
Studies 1A and 1B provide initial evidence that attributing systemic
discrimination to implicit rather than explicit bias results in lower
perceived perpetrator accountability, consistent with past research
(Cameron et al., 2010). Study 1B also revealed lower support for pun-
ishment and concern about the bias when discrimination is attributed to
implicit rather than explicit beliefs. Interestingly, support for reform
efforts to combat the influence of bias in medical interactions did not
differ among participants in the implicit and explicit bias conditions.
4. Study 2
Studies 1A and 1B provide evidence that attributing discrimination
to implicit versus explicit bias reduces perceptions of perpetrator ac-
countability and, perhaps also, support for punishment and concern
about bias and discrimination. The outcomes of the discrimination in
these studies, however, were relatively modest—unsatisfactory doctor-
patient interactions. Reduced accountability for discrimination that is
attributed to implicit bias may not emerge when the outcome is more
harmful. Alternatively, even when harm is severe, people excuse see-
mingly unintentional behaviors (Ames & Fiske, 2013). Thus, implicit
bias may still undermine accountability. Study 2 sought to investigate
this possibility. Specifically, Study 2 sought to discern whether reduced
accountability for implicit, relative to explicit, bias attribution emerges
for discriminatory outcomes that are more harmful than the un-
satisfactory interactions described in Study 1. Study 2 also examined
these questions in a second bias domain; namely, ageism.
Table 1
Correlations of dependent variables for studies 1A, 1B, and 2.
Study 1A Study 1B Study 2
1. 2. 3. 1. 2. 3. 1. 2. 3.
1. Accountability
2. Punishment 0.57 0.56 0.52
3. Concern 0.70 0.64 0.69 0.58 0.72 0.54
4. Reform 0.59 0.66 0.58 0.57 0.62 0.55 0.63 0.58 0.63
All correlations are significant at the p< .01 level.
1
In Study 1A, a scale item referred to “implicit”bias, however, no participant
noticed nor do the results change with its exclusion from the composite. This
was corrected in Study 1B.
2
Analyses in all studies are robust to the inclusion of participant gender, age,
race, and conservatism as covariates.
N.M. Daumeyer, et al. Journal of Experimental Social Psychology 84 (2019) 103812
3
4.1. Method
4.1.1. Participants
Through Amazon Mechanical Turk, 427 United States residents
participated in exchange for $1.00. One person reported suspicion of
the manipulation, leaving a total of 426 participants (53.1% female,
68.5% White, M
age
= 36.28, 209 in the explicit condition; 211 in the
low harm condition).
3
For Study 2, we again sought to enroll a suffi-
cient number of participants to yield an analysis sample with at least
100 people in each condition.
4.1.2. Bias manipulation
As in the previous two studies, participants read an article about
research finding that doctors demonstrated bias in their behavior to-
ward their patients, this time, ageism toward older patients (see, again,
osf.io/5gtyv). In the article, ageism was defined as beliefs associating
older adults with incompetence. Awareness was again the primary
marker of implicit vs. explicit bias. In the explicit condition participants
read, “The doctors were aware that they had these negative beliefs, and
felt they were valid because they were based on actual experiences and
interactions with older adult patients.”In the implicit condition parti-
cipants read, “The doctors had no conscious knowledge of their implicit
ageism nor that their ageism was affecting their treatment of patients
based on age.”In order to ensure participants knew that the doctors
were “well-meaning”in both conditions, each article also reported that:
“Nearly all of the doctors in the study thought that they treated all of
their patients with the same level of care, concern, and attention.”
4.1.3. Harm manipulation
In both conditions, the initial outcomes associated with the doctors'
behavior were similar to those described in Study 1: “Doctors with more
ageist attitudes tended to spend less time with and demonstrate more
dismissive body language toward older patients.”
In the low-harm condition, participants also read, “Furthermore, the
researchers found that older patients of doctors with high levels of age
bias reported being less satisfied with their medical encounter than
patients of doctors with little to no age bias.”In addition, the low-harm
condition article ended with a statement highlighting that “age bias
could be a potential problem for patient care, including leading to
undesirable patient interactions.”
In the high-harm condition participants read, “Furthermore, this
[poor treatment] led to older patients getting less medical attention
than they needed. Of greater concern, the researchers found that older
patients of doctors with high levels of age bias do not live as long as
older patients of doctors with little to no age bias.”In addition, the
high-harm condition article concluded with a statement highlighting
that, “age bias could be a potential problem for patient care, including
leading to premature death.”
4.1.4. Dependent variables
The dependent measures were the same as described in Study 1B,
but modified to refer to ageism rather than political bias. In addition,
the concern scale was increased to seven items in order to include
statements reflecting concern about the bias itself (e.g., “The bias I read
about in the article is concerning”), as in our previous studies, and
concern about the outcomes of the bias (e.g., “The treatment I read
about in the article is concerning”). The items nevertheless formed a
reliable single concern scale (α= 0.87), with higher scores reflecting
greater concern.
4.2. Procedure
After providing informed consent, participants read one of the four
articles (i.e., explicit bias/low-harm; explicit bias/high-harm; implicit
bias/low-harm; implicit bias/high-harm), based on random assignment.
They next completed the primary outcome variables, and provided re-
levant demographic information (e.g., age, race, gender, occupation,
political conservatism). Last, participants answered the data quality and
suspicion check questions, were thanked, debriefed, and paid.
4.3. Results
The dependent variables were thus subjected to a 2 (bias type: ex-
plicit vs. implicit) × 2 (harm: low vs. high) analysis of covariance. The
sample provided 80% power to detect an effect size of η
2
partial
= 0.019.
Given the bias domain featured in this study, however, analyses ex-
ploring moderation by participant age are provided in the supplemental
materials. Correlations among the dependent measures are provided in
Table 1. The means and standard deviations are provided in Table 2.
4.3.1. Accountability
Consistent with predictions and replicating Studies 1A and 1B, the
main effect of bias attribution on accountability was significant, F
(1,422) = 13.66, p< .001, η
2
partial
= 0.031. Unexpectedly, the main
effect of harm was not, F(1,422) = 0.55, p= .459, η
2
partial
= 0.001; and,
importantly, the interaction between bias attribution and harm was also
not significant, F(1,422) = 0.04, p= .837, η
2
partial
< 0.001. That is,
participants who read about implicit bias held the doctors significantly
less accountable than participants who read about explicit bias, re-
gardless of the level of harm associated with the discriminatory beha-
vior.
4.3.2. Punishment
Consistent with Study 1B and as depicted in Fig. 1, participants in
the implicit bias condition supported punishing the doctors less than
participants in the explicit bias condition, F(1,422) = 4.26, p= .040,
η
2
partial
= 0.010. The main effect of harm was also significant, F
(1,422) = 8.48, p= .004, η
2
partial
= 0.020, with high harm resulting in
greater support for punishment than low harm. Again, however, the
interaction between bias attribution and harm was not significant, F
Table 2
Means and Standard Deviations of Dependent Variables by Condition for Studies 1A, 1B, and 2.
Study 1A Study 1B Study 2
Explicit
N= 151
Implicit
N= 116
Explicit
N= 149
Implicit
N= 150
Explicit
N= 209
Implicit
N= 217
MSDMSDMSDMSDMSDM SD
Accountability 5.35
a
0.97 4.88
b
1.17 5.39
a
0.99 4.81
b
1.00 5.50
a
0.93 5.15
b
1.03
Punishment 4.51
a
1.49 4.32
a
1.51 4.83
a
1.54 4.14
b
1.49 4.94
a
1.29 4.68
b
1.37
Concern 4.72
a
1.21 4.56
a
1.18 4.95
a
1.16 4.50
b
1.12 5.16
a
1.07 4.93
b
1.07
Reform 5.06
a
1.14 4.94
a
1.20 5.17
a
1.19 4.98
a
1.15 5.53
a
0.96 5.32
b
1.10
Means with different superscripts within each study and variable differ significantly at the p< .05 level. The means for Study 2 are collapsed across levels of harm.
3
4.0% of participants reported a medical profession (e.g., nurse), however,
their exclusion did not alter the results.
N.M. Daumeyer, et al. Journal of Experimental Social Psychology 84 (2019) 103812
4
(1,422) = 0.61, p= .435, η
2
partial
= 0.001.
4.3.3. Concern
Replicating Study 1B, participants in the implicit bias condition
reported less concern about the bias (and its outcomes) than partici-
pants in the explicit bias condition, F(1,422) = 5.20, p= .023,
η
2
partial
= 0.012. Neither the main effect of harm, F(1,422) = 0.23,
p= .633, η
2
partial
= 0.001, nor the bias attribution by harm interaction,
F(1,422) = 0.61, p= .435, η
2
partial
= 0.001, were significant.
4.3.4. Reform
Unlike in the previous studies, support for reform differed sig-
nificantly as a function of bias attribution, F(1,422) = 4.73, p= .030,
η
2
partial
= 0.011. Participants in the implicit bias condition expressed
less support for reform than participants in the explicit bias condition.
There was no main effect of harm, F(1,422) = 0.52, p= .470,
η
2
partial
= 0.001, nor an interaction between bias attribution condition
and harm, F(1,422) = 0.18, p= .673, η
2
partial
< 0.001.
4.4. Discussion
Study 2 provides additional evidence that when discrimination is
attributed to implicit rather than explicit bias, people hold the perpe-
trators less accountable. Notably, even doctors whose discrimination
was linked to premature patient death (i.e., severe harm) were held less
accountable when the doctors were said to be unaware (implicit con-
dition), rather than aware (explicit condition) of their ageist beliefs.
Replicating Study 1B, Study 2 participants expressed less support for
punishing perpetrators and less concern about the bias in the implicit
compared with explicit bias condition. They also revealed differential
support for reform efforts when the discrimination described was at-
tributed to implicit, rather than explicit, bias. These findings suggest
that the communication of scientific studies detailing the effects of
implicit bias on behavior may unwittingly increase tolerance for both
discriminators and discrimination itself, even when the harm is quite
severe.
5. Study 3
Given our concern about potential unexpected effects of increased
public attention on implicit bias, Study 3 sought to extend our ex-
amination to a domain in which the role of implicit bias in dis-
criminatory or otherwise disparate outcomes has penetrated public
consciousness; namely, police misconduct against racial minority citi-
zens. Specifically, we explored whether attributing racially disparate
outcomes of police-citizen encounters to implicit, rather than explicit,
bias reduces perceptions of police accountability.
Because of the context of the discrimination—namely, socially
sensitive and race-related, we explored whether individual differences
in concern about appearing racially prejudiced might affect judgments
of the police officers either in general or as a function of the implicit vs.
explicit bias attribution. Specifically, participants completed Plant and
Devine's (1998) measures of internal (IMS) and external (EMS) moti-
vation to respond without prejudice. IMS assesses the extent to which
people are intrinsically motivated to behave in non-prejudiced ways
(e.g., to live up to their values), whereas EMS assesses the extent to
which they are externally motivated to respond in non-prejudiced ways
(e.g., to avoid public condemnation). Given that participants made their
responses to the scenarios in the present work anonymously (i.e., on-
line), we did not expect EMS to be relevant to these judgments.
How might IMS moderate judgments of perpetrators who dis-
criminate because of implicit, compared to explicit, racial bias? Past
research has found that IMS can shape reactions to one's own displays
of racial bias, most notably feelings of compunction for failing to reg-
ulate one's implicit bias successfully (Devine, Monteith, Zuwerink, &
Elliot, 1991;Devine, Plant, Amodio, Harmon-Jones, & Vance, 2002;
Legault, Green-Demers, Grant, & Chung, 2007;Plant & Devine, 1998).
When confronted with the possibility that one harbors implicit bias, for
instance, people with high, relative to low, levels of IMS attempt to
reduce their implicit bias through training (Plant & Devine, 2009) and
subsequently behave in ways that are less discriminatory (Cooley, Lei,
& Ellerkamp, 2018). Based on this work, it is possible that individuals
higher (compared with lower) in IMS may not show reduced account-
ability for people who discriminate due to implicit bias relative to those
who discriminate due to explicit bias.
A quite different pattern of results, however, is also possible.
Specifically, one hallmark of being internally motived to respond
without prejudice is harboring low levels of explicit racial bias (Plant &
Devine, 1998), and certainly doing one's best not to behave in dis-
criminatory ways due to any explicitly-held biases (Legault et al., 2007;
Plant & Devine, 2009). Coupled with general societal condemnation of
explicit forms of racial bias (Crandall & Eshleman, 2003), it is possible
that individuals with higher levels of IMS may be especially likely to
hold perpetrators of discrimination that is thought to be due to explicit
bias more accountable and see them as more blameworthy than per-
petrators of implicit bias, whereas individuals lower in IMS may be less
likely to distinguish between individuals who discriminate because of
implicit, compared with explicit, racial bias. Such a pattern would
suggest, further, that individuals with higher levels of IMS are more
sympathetic to perpetrators of racial discrimination that is attributed to
implicit, rather than to explicit, racial bias, perhaps because they can
readily identify with such individuals. The present work examined
these possibilities and, in so doing, is a rare exploration of the extent to
which perceivers' own prejudice-related motivations shape moral
judgments regarding perpetrators of discrimination.
In addition, Study 3 sought to address ambiguity in the perceived
accountability measure. Unlike our previous studies, we assessed per-
ceived accountability for having implicitly (compared to explicitly)
biased beliefs separately from accountability for biased behavior. Biased
beliefs are typically thought to be less controllable and, likely, less
changeable than behavior (Bargh, 1999;Devine, 1989), and largely
outside the domain of moral or legal judgment (Mitchell & Tetlock,
2006). Consequently, perpetrators may assess accountability for im-
plicit vs. explicit attitudes somewhat differently than accountability for
the behaviors they engender. In Study 3, we also offered participants an
opportunity to differentiate between the culpability they assign the
perpetrators themselves and relevant institutional actors. Specifically,
we assessed support for punishment and for reform efforts at both the
individual (i.e., police officer) and institutional (i.e., police department)
levels. It is possible that people hold individual police officers less ac-
countable for discrimination born of implicit, compared with explicit,
bias, but hold police departments equally accountable or, even, hold
departments even more responsible for combatting their officers' im-
plicit compared with explicit bias.
Fig. 1. Bias attribution and harm level on support for punishment. The effect of
bias attribution condition (explicit vs. implicit) and harm condition (low vs.
high) on support for punishment in Study 2. Error bars represent 95% con-
fidence intervals.
N.M. Daumeyer, et al. Journal of Experimental Social Psychology 84 (2019) 103812
5
5.1. Methods
5.1.1. Participants
Through Amazon Mechanical Turk, 227 self-identified White
Americans participated in exchange for $1.00. Four participants were
excluded due to suspicion, resulting in a sample of 223 White partici-
pants (48.4% female, M
age
= 36.81, 111 in explicit condition).
Consistent with our previous studies, we sought to enroll a sample with
at least 100 people in each condition.
5.1.2. Bias manipulation
As in the previous studies, participants read an ostensible news ar-
ticle about a scientific study of a large metropolitan area in which
police officers were found to behave in racially biased ways during
interactions with racial minority citizens. The consequences of the bias
were the same regardless of the bias attribution: police officers with
greater racial bias were said to behave more aggressively toward and
were more likely to handcuffor detain racial minorities compared to
Whites. We again manipulated explicit vs. implicit bias attribution by
emphasizing the unconscious nature of implicit bias.
5.1.3. Dependent measures
All items were rated on Likert scales from 1 (strongly disagree) to 7
(strongly agree).
5.1.3.1. Accountability. Accountability for having biased beliefs was
measured with three items (α= 0.76; e.g., “Police officers should be
held accountable for their biased racial attitudes and beliefs”), as was
accountability for biased behaviors (α= 0.73; e.g., “Police officers
should be held accountable for how they treat racial minority citizens”).
5.1.3.2. Punishment. Support for individual punishment (α= 0.88)
was measured by having participants indicate their agreement with
five punishments following the stem, “Police officers who consistently
demonstrate more negative behavior toward minority, compared with
white, citizens should be…” (taken offpatrol, fined, suspended without
pay, demoted, and fired). Support for institutional-level punishment
(α= 0.83) was measured by having participants indicate their
agreement with four punishments following the stem, “Police
departments with officers who consistently demonstrate more
negative behavior toward minority, compared with white, citizens
should be…” (investigated, fined, forced to hire a new chief, and
taken over by the Justice Department).
5.1.3.3. Reform. Individual reform was measured with three items
(α= 0.85; e.g., “Police officers should be required to undergo
training to prevent any biased attitudes they may have from
impacting their treatment of citizens”). Institutional-level reform was
measured with three items (α=0.79, e.g., “Police departments should
work on developing better relationships with racial minority
communities to reduce the likelihood of negative officer–citizen
interactions”).
5.1.3.4. IMS. Internal motivation to respond without prejudice (IMS;
Plant & Devine, 1998) was assessed with five items (α= 0.91, e.g., “I
am personally motivated by my beliefs to be nonbiased”) rated on 1–7
agreement scales. Because IMS was measured across all studies, we
adapted the scale to measure motivation to respond without prejudice
in general, rather than specifically regarding Black Americans. Higher
scores reflect greater levels of internal motivation.
5.2. Procedure
After providing informed consent, participants read one of the
randomly assigned bias attribution articles. They then completed the
outcome measures, followed by the IMS scale (embedded among a
number of other measures). Participants next reported their demo-
graphic information (e.g., age, gender, race, conservatism, occupation),
followed by data quality, attention, and suspicion checks. After, they
were thanked, debriefed, and paid.
5.3. Results
Preliminary analyses confirmed that IMS scores did not vary sys-
tematically due to the bias attribution manipulation, t(221) = −1.54,
p= .124. To test our predictions, then, we subjected each of the de-
pendent measures to a regression with bias attribution condition (ex-
plicit = −0.5, implicit = 0.5), IMS scores (centered), and the interac-
tion of these two variables as predictors. The analyses had 80% power
to detect an effect size of R
2
partial
= 0.028. Means and standard devia-
tions presented in the text are adjusted for the effects of IMS and the
condition × IMS interaction. The correlations among the variables and
are provided in Table 3.
5.3.1. Accountability
Because the two measures of accountability (for biased beliefs, for
biased behavior) were highly correlated (r= 0.86), we averaged the
items to form a single accountability composite. Analyses revealed a
significant effect of IMS, B= 0.47, SE = 0.04, t(219) = 10.92,
p< .001, with participants higher in IMS holding the police officers
more accountable than participants lower in IMS. The main effect of
bias attribution manipulation was just at the threshold of significance,
B=−0.21, SE = 0.11, t(219) = −1.97, p= .050, and the means were
in the predicted direction. Participants who read that the police officers'
discrimination was due to implicit bias (M= 5.44, SD = 0.81) held
them less accountable than participants who read that the police offi-
cers' discrimination was due to their explicit bias (M= 5.65,
SD = 0.81). This effect was not moderated by IMS, B=−0.08,
SE = 0.09, t(219) = −1.01, p= .314.
5.3.2. Punishment
Analyses of support for individual-level punishment, again, revealed
a significant effect of IMS, B= 0.46, SE = 0.07, t(219) = 6.84,
p< .001; higher levels of IMS predicted greater support for punishing
the police officers. Although the means were in the predicted direction,
the main effect of bias attribution was not statistically reliable,
B=−0.29, SE =0.17, t(219) = −1.75, p= .082, nor was the bias
attribution by IMS interaction, B=−0.13, SE = 0.13, t(219) = −0.98,
p= .327. Participants who read that police officer's discrimination was
due to implicit bias (M= 4.87, SD = 1.26) did not significantly differ in
their support for punishing individual policers from participants who
read their discrimination was due to explicit bias (M= 5.17,
SD = 1.26).
Support for institutional-level punishment (i.e., police departments)
was subjected to the same analysis. Once again, the main effect of IMS
emerged, with those higher in IMS again expressing greater support for
punishing police departments, B= 0.46, SE = 0.06, t(219) = 7.48,
p< .001. Contrary to predictions, the main effect of bias attribution
Table 3
Correlations of variables for study 3.
Study 3
1. 2. 3. 4. 5.
1. Accountability
2. Individual Punishment 0.64
3. Departmental Punishment 0.65 0.79
4. Individual Reform 0.70 0.53 0.59
5. Departmental Reform 0.66 0.43 0.53 0.79
6. IMS 0.59 0.42 0.46 0.59 0.53
All correlations were significant at the p< .01 level.
N.M. Daumeyer, et al. Journal of Experimental Social Psychology 84 (2019) 103812
6
condition was not reliable, B=−0.12, SE = 0.16, t(219) = −0.75,
p= .453, however, the IMS by condition interaction did reach con-
ventional levels of statistical significance, B=−0.27, SE = 0.12, t
(219) = −2.15, p= .033.
To decompose this interaction, we used PROCESS (model 1) to test
for the effect of bias attribution condition at low (−1 SD from the
mean) and high (+1 SD from the mean) levels of internal motivation to
respond without prejudice (IMS). Analyses revealed that among those
low in IMS, the effect of bias attribution condition was not significant,
B= 0.22, SE = 0.22, t(219) = 0.99, p= .322. Participants with higher
levels of IMS, however, did differentiate between department-level
punishment after reading about officer discrimination due to explicit
compared with implicit bias, B=−0.46, SE = 0.22, t(219) = −2.07,
p= .039; as depicted in Fig. 2, high IMS participants expressed greater
support for punishing police departments when they read about officers
discriminating due to explicit rather than implicit bias. For the full
model see supplemental materials.
5.3.3. Reform
Analyses of participants' support for individual-level reform efforts
for reducing police officer discriminatory behavior revealed a pattern
similar to that found for departmental punishment. Specifically, al-
though the effect of bias attribution was not significant, B=−0.13,
SE = 0.12, t(219) = −1.02, p=.309, a significant main effect of IMS
emerged, B= 0.52, SE = 0.49, t(219) = 10.55, p< .001, which was
qualified by a significant interaction with bias attribution condition,
B=−0.22, SE = 0.10, t(219) = −2.19, p= .030. We again used
PROCESS (model 1) to decompose this interaction, examining the effect
of bias attribution at low (−1 SD) and high (+1SD) levels of IMS.
Analyses again revealed that among participants low in IMS, the effect
of bias condition was not significant, B= 0.15, SE = 0.18, t
(219) = 0.83, p= .405. However, among participants high in IMS, the
effect of bias attribution condition was significant, B=−0.40,
SE = 0.178 t(219) = −2.29, p= .023. Participants high in internal
motivation to respond without prejudice supported reform efforts tar-
geted at individual officers more when discrimination was attributed to
explicit, rather than implicit, bias. Again, for the full model see sup-
plemental materials.
Analyses of support for institutional-level reform efforts targeted at
police departments revealed only a main effect of IMS, B= 0.40,
SE = 0.04, t(219) = 9.02, p< .001. Regardless of bias attribution
condition, participants higher in IMS were more supportive of depart-
ment-level reform efforts than participants lower in IMS. Neither the
bias attribution manipulation [B=−0.09, SE =0.11, t(219) = −0.84,
p= .401] nor its interaction with IMS [B=−0.06, SE = 0.09, t
(219) = −0.72, p= .471] were reliable. Participants who read that
police officers discriminate because of implicit bias (M= 5.69,
SD = 1.18) did not significantly differ from participants who read that
police officers discriminate because of explicit bias (M= 5.43,
SD = 1.37) in their support for reform efforts aimed at police depart-
ments.
5.4. Discussion
In a domain that has received considerable public attention—ra-
cially disparate patterns of police misconduct—Study 3 revealed that
attributing discrimination to implicit rather than explicit bias reduces
perceived perpetrator accountability. Further, individual differences in
internal motivation to respond without prejudice (IMS) predicted
overall assessments of perpetrator accountability, support for punish-
ment, and support for reform. Specifically, high IMS participants tended
to hold the police officers more accountable for discriminatory behavior
than low IMS participants, irrespective of the bias attribution. For in-
stitutional-level punishment and individual-level reform, interestingly
it was the high IMS participants who differentiated between dis-
crimination based in implicit and explicit racial bias. That is, partici-
pants with higher levels of IMS supported more severely punishing
police departments and more reform efforts for individual officers when
police officer bias was said to be explicit, rather than, implicit.
Participants lower in IMS did not reveal these differences. Taken to-
gether, these findings suggest that judgments beyond perpetrator ac-
countability for discrimination due to implicit vs. explicit bias may be
shaped, at least in part, by domain-relevant motivations of perceivers.
Future research is, of course, necessary to replicate and explore this
finding more deeply.
6. Internal meta-analysis
In order to assess the robustness of the effects observed across our
studies, we conducted an internal meta-analysis (Goh, Hall, &
Rosenthal, 2016). Specifically, we calculated Cohen's dfor the explicit
vs. implicit bias attribution effect using the means and standard de-
viations for each dependent variable in each study. For Study 2, we
collapsed across harm to generate the estimates. For Study 3, we cre-
ated composites for punishment and reform that collapsed across the
individual- and institutional-level distinction, and then re-calculated
the bias attribution manipulation effect-size estimates for each. Given
there were no significant main effects of bias attribution on punishment
and reform in Study 3, this decision should provide conservative esti-
mates of each effect-size. We also used the adjusted means and standard
deviations (controlling for IMS and the condition × IMS interaction) to
calculate Cohen's d. Analyses revealed a significant effect of account-
ability (d= 0.41, Z= 7.08, p< .001). Across all studies, when parti-
cipants read that discriminatory behavior was due to implicit compared
to explicit bias, they perceived less perpetrator accountability (see
Fig. 3). Additionally, analyses revealed modest, albeit statistically sig-
nificant, overall effect-size estimates for each of the other outcome
Fig. 2. Support for institutional punishment by bias attribution condition and
internal motivation to respond without prejudice (IMS). High and low IMS
represent +/−1 SD from the mean, respectively.
Fig. 3. Effect sizes for accountability across studies. Cohen's dfor accountability
plotted for each study. Error bars represent 95% confidence intervals.
N.M. Daumeyer, et al. Journal of Experimental Social Psychology 84 (2019) 103812
7
variables: punishment (d= 0.24, Z= 4.20, p< .001), concern
(d= 0.25, Z= 3.90, p< .001), and reform (d= 0.16, Z= 2.81,
p= .005).
7. General discussion
Scholars of stereotyping and prejudice have long expressed concern
that emphasizing the role of unconscious, automatic beliefs in en-
gendering discrimination might reduce the perceived culpability of its
perpetrators (Fiske, 2004). The present findings suggest this concern
was well-founded. Specifically, in the context of scientific research
communications, we found that people hold perpetrators less accoun-
table for discriminatory behavior when it is attributed to their implicit,
rather than explicit, attitudes. This reduced accountability effect was
observed in two contexts (medical and police interactions), across three
different biases (political, age-based, racial), and, somewhat surpris-
ingly, was not attenuated when the consequences of the discrimination
were especially severe (i.e., premature death). Given that a great deal of
discrimination is rooted in implicit forms of bias (Greenwald &
Pettigrew, 2014), the small, but reliable, reduced accountability effect
found here could be highly consequential.
In addition to being held less accountable for discrimination, per-
petrators were somewhat less likely to be punished if their behavior was
described as stemming from implicit rather than explicit bias.
Moreover, perceivers expressed lower levels of concern about, and were
less likely to support efforts to mitigate against the effects of, implicit
compared with explicit bias. Taken together, these findings suggest that
communications of scientificfindings regarding the effects of implicit
bias may unwittingly reduce the perceived severity of the discrimina-
tion it engenders.
7.1. Theoretical implications
The primary finding of the present work—reduced accountability
for discriminatory acts born of implicit rather than explicit bias—is
consistent with prevailing models of moral reasoning, all of which
underscore the essential role of a person's mental state in shaping
judgments of culpability for harmful actions (Cushman, 2015;Knobe &
Nichols, 2011;Malle et al., 2014; see also Heider, 1958). Most notably,
an actor needs to have some level of awareness of and control over their
behavior in order to be held morally responsible for it (e.g., Alicke,
2000;Shaver, 1985). Hence, as we observed here, people judge per-
petrators of harmful actions more negatively when they are thought to
have engaged in those acts knowingly, consciously, and/or in-
tentionally (Monroe & Malle, 2017, 2019; see also Cameron et al., 2010;
Redford & Ratliff, 2016). In the present work, that is, perpetrators of
discrimination who were said to be influenced by attitudes and beliefs
that they held unconsciously (i.e., implicitly) were held less accoun-
table than perpetrators of the same discriminatory behavior who were
said to be influenced by attitudes and beliefs that they held consciously
(i.e., explicitly).
While our results may not be surprising based on this larger theo-
retical work, they are particularly compelling given that the behaviors
examined across these studies are far more complex than the types of
behaviors typically studied in the moral reasoning literature.
Specifically, relatively rich descriptions of several behaviors are pro-
vided in the articles used in the present studies rather than exposing
participants to discrete behaviors that target individuals ostensibly
have engaged in. In addition, many of the behaviors described in our
work can surely be classified as conscious (or even intentional) acts,
including aggressive behavior toward citizens by police officers. That is,
on their own, they could and perhaps should be understood as deser-
ving of the level of moral condemnation associated with intentional
behavior. Yet, describing the differential behaviors observed as stem-
ming from implicit, rather than explicit, bias seems to offer perceivers
an adequate basis for reducing judgments of accountability.
Should the tenets of prevailing models of moral responsibility apply
to these types of behaviors? Imagine that a person regularly and re-
peatedly steals the morning newspaper from one of his neighbors.
Would his perceived accountability for this action be reduced if we later
find out that he is unaware (vs. aware) of his negative affect toward the
neighbors? It certainly seems unlikely. Similarly, the doctors and police
officers described in our work knowingly engaged in at least some of
the behaviors described in the scenarios (e.g., aggressive body lan-
guage). What they are said to be unaware of is the extent to which they
were behaving differently toward different types of targets based on
their social group memberships. That is, the relation between their
attitudes regarding and behaviors toward members of a disfavored,
relative to favored, social category. The extent to which current models
of moral reasoning should be relevant to these behaviors is not clear,
but the present findings suggest that perceivers lower their assessments
of accountability nonetheless. Future research is needed to discern ex-
actly when, why, and for whom accountability for even clearly inten-
tional behaviors is reduced by locating a possible origin of the behavior
to attitudes and beliefs that are held implicitly rather than explicitly
(see Monroe & Malle, 2017, 2019).
7.2. Practical implications
At first glance, one potential takeaway from the present work is for
researchers to avoid discussing implicit bias with the public. This is
certainly not our position. Indeed, we maintain that the penetration of
the implicit bias construct into public consciousness has certainly been
a much-needed corrective to outdated models of stereotyping, pre-
judice, and discrimination that require clear evidence of antipathy,
animus, and often discriminatory intent in order to classify behavior as
biased (Allport, 1954;Simon, Moss, & O'Brien, 2019;Sommers &
Norton, 2006;Swim, Scott, Sechrist, Campbell, & Stangor, 2003; see
also Washington v. Davis, 1976). Rather than calling for reduced public
discussion of the science of implicit bias, we believe it is time for more
nuanced public conversations. For example, it may be time to revisit the
tendency for researchers and others to describe implicit bias as un-
conscious and/or uncontrollable. Not only is there evidence that people
do have some ability to detect their implicit biases (Hahn, Judd, Hirsh,
& Blair, 2014;Monteith, Voils, & Ashburn-Nardo, 2001;Uhlmann &
Nosek, 2012), but the present results suggest that implicit bias attri-
butions—that is, the lack of awareness of bias—engender reduced ac-
countability judgments. In addition, public conversations about implicit
bias could also include information about the potential for individuals
to override the effects of even implicitly-held attitudes and beliefs on
their behavior, at least when they have the opportunity and motivation
(Dunton & Fazio, 1997;Nosek, Hawkins, & Frazier, 2011). Public
conversations should also highlight the potential for policies and de-
cision-making structures (e.g., diverse hiring panels) to combat the
influence of implicit bias on individual acts and judgments (e.g.,
Daumeyer, Rucker, & Richeson, 2017).
Relatedly, it may be possible to capitalize on the popular knowledge
of implicit bias in order to help combat it. At some point, awareness
that implicit bias is a common pathway toward the reproduction of
unequal and unjust societal outcomes based on race, gender, age, and
other classifications, should motivate efforts to combat it structurally
and institutionally, in addition to any individual-level efforts (see Kelly
& Roedder, 2008). In addition, the more widely-known implicit bias
becomes, the more people (and relevant institutions) can and should be
held accountable for its effects. That is, when people think that actors
should have been able to foresee the harmful consequences of even
their unintended actions, they hold those actors more accountable
(Lagnado & Channon, 2008;Laurent, Nuñez, & Schweitzer, 2016;
Monroe & Malle, 2019). Thus, rather than calling for less public dis-
cussion of implicit bias, we argue for a more sophisticated conversation
on the ways in which implicit bias shapes behavior and multiple ways
to combat it (e.g., Payne, Vuletich, & Lundberg, 2017).
N.M. Daumeyer, et al. Journal of Experimental Social Psychology 84 (2019) 103812
8
7.3. Limitations
One limitation of the present work is its failure to probe whether the
results may differ for people who share a meaningful group member-
ship with the victims of the discrimination. There is reason to expect
that the targets of the discrimination may not hold perpetrators any less
accountable for discrimination based on an implicit, rather than ex-
plicit, bias attribution. Not only would such a finding reveal an im-
portant moderator of the accountability effect found here, but it would
also suggest that motivation may play a role in engendering what has
heretofore largely been understood as a purely reasoned discounting of
moral responsibility. The patterns of moderation by internal motivation
to respond without prejudice that emerged in Study 3 suggest a role for
motivation in shaping these judgments, this question is especially ripe
for examination in future research.
It is also important to acknowledge that the primary manipulation
in the present work distinguished implicit from explicit bias by ma-
nipulating awareness. While this decision is grounded in and, thus,
consistent with past work (e.g., Redford & Ratliff, 2016), and as noted
previously, reflects how researchers tend to discuss implicit bias both in
the literature (Greenwald & Banaji, 1995) and with the public (Payne,
Niemi, & Doris, 2018), it is only one of the characteristics that differ-
entiates implicit from explicit bias. Had we focused on other relevant
distinguishing characteristics (e.g., controllability, efficiency, auto-
maticity, etc.), it is possible that a different pattern of results may have
emerged. For example, “automatic”bias that is described as potentially
controllable does not yield reduced moral responsibility judgments re-
lative to explicit bias (Cameron et al., 2010).
Further, discrimination born of implicit bias is sometimes thought to
be unintentional or at least less intentional than discrimination born of
explicit bias (Cameron et al., 2010;Onyeador, 2017). Thus, had we
defined implicit bias by a lack of intentionality, rather than awareness,
we would expect similar results to the ones found here. Indeed, they
may have been even more robust. Clarifying the role of perceived intent
as a potential mediator between holding implicit forms of bias and
reduced accountability for behaviors that are said to stem from such
bias is an essential direction for future research. This issue is especially
important, given the relevance of intent to legal definitions of inter-
personal discrimination (Washington v. Davis, 1976). More research is
needed to understand how beliefs about the awareness of bias, con-
trollability of its influence on behavior, and intention to engage in
specific behaviors that turn out to be discriminatory shape perceptions
of accountability for discrimination.
7.4. Conclusion
The present work revealed that professionals such as doctors and
police officers are held less accountable for discriminatory behavior
born of implicit, compared with explicit, bias. Moreover, this reduction
in accountability was not accompanied by an increase in support for
efforts to reduce bias through trainings or other institutional policies.
Not only does this work have important implications for how re-
searchers communicate the science of implicit bias, but it begs the
question, can the discrimination born of implicit bias be reduced if no
one is held responsible for it?
Open practices
The studies presented in this paper have earned Open Materials and
Open Data badges for transparent practices. All data and materials for
the studies presented here can be found at https://osf.io/5gtyv/.
Acknowledgements
This research was supported by an NSF Social, Behavioral, and
Economic Sciences (SBE) Postdoctoral Research Fellowship
(#1809370) awarded to the second author.
Appendix A. Supplementary data
Supplementary data to this article can be found online at https://
doi.org/10.1016/j.jesp.2019.04.010.
References
Alicke, M. D. (2000). Culpable control and the psychology of blame. Psychological Bulletin,
126, 556–574. https://doi.org/10.1037//0033-2909.126.4.556.
Allport, G., W. (1954). The nature of prejudice.Oxford, England: Addison-Wesley.
Ames, D. L., & Fiske, S. T. (2013). Intentional harms are worse, even when they're not.
Psychological Science, 24, 1755–1762. https://doi.org/10.1177/0956797613480507.
Axt, J. R., Ebersole, C. R., & Nosek, B. A. (2014). The rules of implicit evaluation by race,
religion, and age. Psychological Science, 25, 1804–1815. https://doi.org/10.1177/
0956797614543801.
Bargh, J. A. (1999). The cognitive monster: The case against the controllability of auto-
matic stereotype effects. In S. Chaiken, & Y. Trope (Eds.). Dual-process theories in social
psychology (pp. 361–383). New York, NY, US: Guilford Press.
Bertrand, M., & Mullainathan, S. (2004). Are Emily and Greg more employable than
Lakisha and Jamal? A field experiment on labor market discrimination. American
Economic Review, 94, 991–1013.
Cameron, C. D., Payne, B. K., & Knobe, J. (2010). Do theories of implicit race bias change
moral judgments? Social Justice Research, 23, 272–289. https://doi.org/10.1007/
s11211-010-0118-z.
Carter, E. R., & Murphy, M. C. (2015). Group-based differences in perceptions of racism:
What counts, to whom, and why? Social and Personality Psychology Compass, 9,
269–280. https://doi.org/10.1111/spc3.12181.
Casad, B. J., Flores, A. J., & Didway, J. D. (2013). Using the implicit association test as an
unconsciousness raising tool in psychology. Teaching of Psychology, 40, 118–123.
https://doi.org/10.1177/0098628312475031.
Clack, Z. (Writer), Rhimes, S. (Credited Writer) & Sullivan, K. R. (Director). (January, 25,
2018). Personal Jesus [Grey's Anatomy]. In D. Allen (Executive Producer). B. Beers
(Executive Producer). Z. Clack (Executive Producer). S. Collins (Supervising
Producer). F. Einesman (Executive Producer). E. Finch (Co-Executive Producer). J.
Ganzel (Co-Executive Producer). M. Gordon (Executive Producer). W. Harper
(Executive Producer). M. Hope (Co-Producer). L. Klien (Supervising Producer). M.
Marinis (Co-Executive Producer). E. Pompeo (Producer). A. Reaser (Co-Executive
Producer). S. Rhimes (Creator). L. Taylor (Producer). K. Veroff(Executive Producer).
S. E. White (Producer). Los Angeles, California: ABC Studios.
Cooley, E., Lei, R. F., & Ellerkamp, T. (2018). The mixed outcomes of taking ownership for
implicit racial biases. Personality and Social Psychology Bulletin, 44, 1424–1434.
https://doi.org/10.1177/0146167218769646.
Crandall, C. S., & Eshleman, A. (2003). A justification-suppression model of the expres-
sion and experience of prejudice. Psychological Bulletin, 129, 414–446. https://doi.
org/10.1037/0033-2909.129.3.414.
Cushman, F. (2015). Deconstructing intent to reconstruct morality. Current Opinion in
Psychology, 6,97–103. https://doi.org/10.1016/j.copsyc.2015.06.003.
Dasgupta, N. (2004). Implicit ingroup favoritism, outgroup favoritism, and their beha-
vioral manifestations. Social Justice Research, 17, 143–169. https://doi.org/10.1023/
B:SORE.0000027407.70241.15.
Daumeyer, N. M., Rucker, J. M., & Richeson, J. A. (2017). Thinking structurally about
implicit bias: Some peril, lots of promise. Psychological Inquiry, 28, 258–261. https://
doi.org/10.1080/1047840X.2017.1373556.
Devine, P. G. (1989). Stereotypes and prejudice: Their automatic and controlled com-
ponents. Journal of Personality and Social Psychology, 56,5–18. https://doi.org/10.
1037/0022-3514.56.1.5.
Devine, P. G., Monteith, M. J., Zuwerink, J. R., & Elliot, A. J. (1991). Prejudice with and
without compunction. Journal of Personality and Social Psychology, 60, 817–830.
https://doi.org/10.1037/0022-3514.60.6.817.
Devine, P. G., Plant, E. A., Amodio, D. M., Harmon-Jones, E., & Vance, S. L. (2002). The
regulation of explicit and implicit race bias: The role of motivations to respond
without prejudice. Journal of Personality and Social Psychology, 82, 835–848. https://
doi.org/10.1037//0022-3514.82.5.835.
Dovidio, J. F., & Gaertner, S. L. (2010). Intergroup bias. (Handbook of Social Psychology).
Dovidio, J. F., Kawakami, K., & Gaertner, S. L. (2002). Implicit and explicit prejudice and
interracial interaction. Journal of Personality and Social Psychology, 82,62–68.
https://doi.org/10.1037//0022-3514.82.1.62.
Dunton, B. C., & Fazio, R. H. (1997). An individual difference measure of motivation to
control prejudiced reactions. Personality and Social Psychology Bulletin, 23, 316–326.
https://doi.org/10.1177/0146167297233009.
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. https://doi.org/10.1037/0022-3514.87.6.876.
Fiske, S. T. (2004). Intent and ordinary bias: Unintended thought and social motivation
create casual prejudice. Social Justice Research, 17, 117–127. https://doi.org/10.
1023/B:SORE.0000027405.94966.23.
Gawronski, B., & Bodenhausen, G. V. (2006). Associative and propositional processes in
evaluation: An integrative review of implicit and explicit attitude change.
Psychological Bulletin, 132, 692–731. https://doi.org/10.1037/0033-2909.132.5.692.
Goh, J. X., Hall, J. A., & Rosenthal, R. (2016). Mini meta-analysis of your own studies:
Some arguments on why and a primer on how. Social and Personality Psychology
N.M. Daumeyer, et al. Journal of Experimental Social Psychology 84 (2019) 103812
9
Compass, 10, 535–549. https://doi.org/10.1111/spc3.12267.
Greenwald, A. G., & Banaji, M. R. (1995). Implicit social cognition: Attitudes, self-esteem,
and stereotypes. Psychological Review, 102,4–27.
Greenwald, A. G., & Pettigrew, T. F. (2014). With malice toward none and charity for
some: Ingroup favoritism enables discrimination. American Psychologist, 69, 669–684.
https://doi.org/10.1037/a0036056.
Hahn, A., Judd, C. M., Hirsh, H. K., & Blair, I. V. (2014). Awareness of implicit attitudes.
Journal of Experimental Psychology. General, 143, 1369–1392. https://doi.org/10.
1037/a0035028.
Heider, F. (1958). The other person as perceiver. The psychology of interpersonal relations
(pp. 59–78). Hillsdale, New Jersey: Lawrence Erlbaum Associates.
Hersh, E. D., & Goldenberg, M. N. (2016). Democratic and republican physicians provide
different care on politicized health issues. Proceedings of the National Academy of
Sciences,113, 11811–11816. https://doi.org/10.1073/pnas.1606609113.
Hester, N., & Gray, K. (2018). For black men, being tall increases threat stereotyping and
police stops. Proceedings of the National Academy of Sciences, 115, 2711–2715. https://
doi.org/10.1073/pnas.1714454115.
Kelly, D., & Roedder, E. (2008). Racial cognition and the ethics of implicit bias. Philosophy
Compass, 3, 522–540. https://doi.org/10.1111/j.1747-9991.2008.00138.x.
Knobe, J., & Nichols, S. (2011). Free will and the bounds of the self. In Oxford Handbook of
Free Will (2nd ed., pp. 530–554). New York: Oxford University Press. https://doi.org/
10.1093/oxfordhb/9780195399691.003.0028.
Lagnado, D. A., & Channon, S. (2008). Judgments of cause and blame: The effects of
intentionality and foreseeability. Cognition, 108, 754–770. https://doi.org/10.1016/j.
cognition.2008.06.009.
Laurent, S. M., Nuñez, N. L., & Schweitzer, K. A. (2016). Unintended, but still blame-
worthy: The roles of awareness, desire, and anger in negligence, restitution, and
punishment. Cognition & Emotion, 30, 1271–1288. https://doi.org/10.1080/
02699931.2015.1058242.
Legault, L., Green-Demers, I., Grant, P., & Chung, J. (2007). On the self-regulation of
implicit and explicit prejudice: A self-determination theory perspective. Personality
and Social Psychology Bulletin, 33, 732–749. https://doi.org/10.1177/
0146167206298564.
Litman, L., Robinson, J., & Abberbock, T. (2017). TurkPrime.com: A versatile crowd-
sourcing data acquisition platform for behavioral sciences. Behavior Research Methods,
49, 433–442. https://doi.org/10.3758/s13428-016-0727-z.
Malle, B. F. (1999). How people explain behavior: A new theoretical framework.
Personality and Social Psychology Review, 3,23–48.
Malle, B. F., Guglielmo, S., & Monroe, A. E. (2014). A theory of blame. The Journal of
Psychology, 25, 147–186. https://doi.org/10.1080/1047840X.2014.877340.
Mitchell, G., & Tetlock, P. E. (2006). Antidiscrimination law and the perils of mind-
reading. Ohio State Law Journal, 67, 1023–1122.
Monroe, A. E., & Malle, B. F. (2017). Two paths to blame: Intentionality directs moral
information processing along two distinct tracks. Journal of Experimental Psychology:
General, 146, 123–133. https://doi.org/10.1037/xge0000234.
Monroe, A. E., & Malle, B. F. (2019). People systematically update moral judgments of
blame. Journal of Personality and Social Psychology, 116, 215–236. https://doi.org/10.
1037/pspa0000137.
Monteith, M. J., Voils, C. I., & Ashburn-Nardo, L. (2001). Taking a look underground:
Detecting, interpreting, and reacting to implicit racial biases. Social Cognition, 19,
395–417. https://doi.org/10.1521/soco.19.4.395.20759.
Nadler, J., & McDonnell, M.-H. (2012). Moral character, motive, and the psychology of
blame. Cornell Law Review, 97, 255–304.
Nosek, B. A., Hawkins, C. B., & Frazier, R. S. (2011). Implicit social cognition: From
measures to mechanisms. Trends in Cognitive Science, 15, 152–159.
Onyeador, I. N. (2017). Presumed unintentional: The ironic effects of implicit bias framing on
Whites' perceptions of discrimination (unpublished doctoral dissertation). Los Angeles,
CA: University of California.
Payne, B. K., Vuletich, H. A., & Lundberg, K. B. (2017). The bias of crowds: How implicit
bias bridges personal and systemic prejudice. Psychological Inquiry, 28, 233–248.
https://doi.org/10.1080/1047840X.2017.1335568.
Payne, K., Niemi, L., & Doris, J. M. (2018). How to think about “implicit bias.”.Scientific
American. Retrieved from https://www.scientificamerican.com/article/how-to-think-
about-implicit-bias/.
Perry, S. P., Murphy, M. C., & Dovidio, J. F. (2015). Modern prejudice: Subtle, but un-
conscious? The role of bias awareness in Whites' perceptions of personal and others'
biases. Journal of Experimental Social Psychology, 61,64–78. https://doi.org/10.1016/
j.jesp.2015.06.007.
Plant, E. A., & Devine, P. G. (1998). Internal and external motivation to respond without
prejudice. Journal of Personality and Social Psychology, 75, 811–832.
Plant, E. A., & Devine, P. G. (2009). The active control of prejudice: Unpacking the in-
tentions guiding control efforts. Journal of Personality and Social Psychology, 96,
640–652. https://doi.org/10.1037/a0012960.
Redford, L., & Ratliff, K. A. (2016). Perceived moral responsibility for attitude-based
discrimination. British Journal of Social Psychology, 55, 279–296. https://doi.org/10.
1111/bjso.12123.
Shaver, K. G. (1985). The attribution of blame: Causality, responsibility, and blameworthiness.
New York, NY: Springer-Verlag.
Simon, S., Moss, A. J., & O'Brien, L. T. (2019). Pick your perspective: Racial group
membership and judgments of intent, harm, and discrimination. Group Processes &
Intergroup Relations, 22, 215–232. https://doi.org/10.1177/1368430217735576.
Sommers, S. R., & Norton, M. I. (2006). Lay theories about white racists: What constitutes
racism (and what doesn't). Group Processes & Intergroup Relations, 9, 117–138. https://
doi.org/10.1177/1368430206059881.
Swim, J. K., Scott, E. D., Sechrist, G. B., Campbell, B., & Stangor, C. (2003). The role of
intent and harm in judgments of prejudice and discrimination. Journal of Personality
and Social Psychology, 84, 944–959. https://doi.org/10.1037/0022-3514.84.5.944.
Uhlmann, E. L., & Nosek, B. A. (2012). My culture made me do it: Lay theories of re-
sponsibility for automatic prejudice. Social Psychology, 43, 108–113. https://doi.org/
10.1027/1864-9335/a000089.
Washington v. Davis, 426 U.S. (1976).
Weiner, B. (1995). Judgments of responsibility: A foundation for a theory of social conduct.
New York: The Guilford Press.
N.M. Daumeyer, et al. Journal of Experimental Social Psychology 84 (2019) 103812
10
