ArticlePDF Available

Long-term reduction in implicit race bias: A prejudice habit-breaking intervention

Authors:

Abstract and Figures

We developed a multi-faceted prejudice habit-breaking intervention to produce long-term reductions in implicit race bias. The intervention is based on the premise that implicit bias is like a habit that can be reduced through a combination of awareness of implicit bias, concern about the effects of that bias, and the application of strategies to reduce bias. In a 12-week longitudinal study, people who received the intervention showed dramatic reductions in implicit race bias. People who were concerned about discrimination or who reported using the strategies showed the greatest reductions. The intervention also led to increases in concern about discrimination and personal awareness of bias over the duration of the study. People in the control group showed none of the above effects. Our results raise the hope of reducing persistent and unintentional forms of discrimination that arise from implicit bias.
Content may be subject to copyright.
Long-term reduction in implicit race bias: A prejudice habit-breaking intervention
Patricia G. Devine , Patrick S. Forscher, Anthony J. Austin
1
, William T.L. Cox
Psychology Department, University of Wisconsin, Madison, WI, USA
HIGHLIGHTS
We developed an intervention to produce long-term reductions in implicit race bias.
The intervention produced reductions in implicit bias that lasted up to 8 weeks.
The intervention also increased awareness of bias and concern about discrimination.
Our results raise the hope of reducing the pernicious effects of implicit race bias.
abstractarticle info
Article history:
Received 28 March 2012
Revised 11 June 2012
Available online 20 July 2012
Keywords:
Prejudice
Stereotyping
Intervention
Reduction
Implicit bias
Self-regulation
We developed a multi-faceted prejudice habit-breaking intervention to produce long-term reductions in im-
plicit race bias. The intervention is based on the premise that implicit bias is like a habit that can be broken
through a combination of awareness of implicit bias, concern about the effects of that bias, and the applica-
tion of strategies to reduce bias. In a 12-week longitudinal study, people who received the intervention
showed dramatic reductions in implicit race bias. People who were concerned about discrimination or who
reported using the strategies showed the greatest reductions. The intervention also led to increases in con-
cern about discrimination and personal awareness of bias over the duration of the study. People in the control
group showed none of the above effects. Our results raise the hope of reducing persistent and unintentional
forms of discrimination that arise from implicit bias.
© 2012 Elsevier Inc. All rights reserved.
Introduction
Despite encouraging trends suggesting that racial prejudice in the U. S.
has waned in the last half century (Gaertner & Dovidio, 1986; Schuman,
Steeh, Bobo, & Krysan, 1997), widespread evidence suggests that Black
people face continuing discrimination and have more adverse outcomes
than White people across a variety of domains related to success and
well-being (e.g., Bertrand & Mullainathan, 2004; Bradford, Newkirk, &
Holden, 2009; Mitchell, Haw, Pfeifer, & Meissner, 2005; Steele, 1997;
Vontress, Woodland, & Epp, 2007). The paradox of persistent racial in-
equalities amid improving racial attitudes has led to a search for factors
underlying ongoing discrimination. Several theorists have implicated
implicit race biases, which are automatically activated and often
unintentional, as major contributors to the perpetuation of discrimina-
tion (e.g., Devine, 1989; Fiske, 1998; Gaertner & Dovidio, 1986).
Supporting this claim, accumulating evidence reveals that implicit
biases are linked to discriminatory outcomes ranging from the seeming-
ly mundane, such as poorer quality interactions (McConnell & Leibold,
2001), to the undeniably consequential, such as constrained employ-
ment opportunities (Bertrand & Mullainathan, 2004) and a decreased
likelihood of receiving life-saving emergency medical treatments
(Green et al., 2007). Many theorists argue that implicit biases persist
and are powerful determinants of behavior precisely because people
lack personal awareness of them and they can occur despite conscious
nonprejudiced attitudes or intentions (Bargh, 1999; Devine, 1989;
Gaertner & Dovidio, 1986). This process leads people to be unwittingly
complicit in the perpetuation of discrimination.
The reality of lingering racial disparities, combined with the empirical-
ly established links between implicit bias and pernicious discriminatory
outcomes, has led to a clarion call for strategies to reduce these biases
(Fiske, 1998; Smedley, Stith, & Nelson, 2003). In response, the eld has
witnessed an explosion of empirical efforts to reduce implicit biases
(Blair, 2002). These efforts have yielded a number of easy-to-implement
strategies, such as taking the perspective of stigmatized others
(Galinsky & Moskowitz, 2000) and imagining counter-stereotypic exam-
ples (Blair, Ma, & Lenton, 2001; Dasgupta & Greenwald, 2001), that lead to
Journal of Experimental Social Psychology 48 (2012) 12671278
We thank Markus Brauer and Carlie Allison for their comments on a previous ver-
sion of this paper. We also thank Becky McGill, Rachel Nitzarim, Julia Salomon, and
Chelsea Wenzlaff for their help in running the experiment reported in this paper. Prep-
aration of this article was supported by NIH grant R01 GM088477.
Corresponding author at: Psychology Department, University of WisconsinMadison,
1202 W Johnson St, Madison, WI 53706, USA.
E-mail address: pgdevine@wisc.edu (P.G. Devine).
1
Now at the University of Chicago, Chicago, IL, USA.
0022-1031/$ see front matter © 2012 Elsevier Inc. All rights reserved.
doi:10.1016/j.jesp.2012.06.003
Contents lists available at SciVerse ScienceDirect
Journal of Experimental Social Psychology
journal homepage: www.elsevier.com/locate/jesp
substantial reductions in implicit bias,atleastforashorttime(i.e.,upto
24 hours). These strategies yield reductions in implicit bias even though
people use the strategies at the experimenter's behest, with no intention
to reduce implicit bias. It is unclear, however, whether such incidental re-
ductions in implicit bias are enduring or whether people could intention-
ally implement such strategies in the service of a long-term goal to reduce
implicit bias.
Although there is no direct evidence about whether one-shot strat-
egies used at another's behest could produce enduring change, some
general dual-process theories in psychology (e.g., Epstein, 1994; Smith
& DeCoster,2000; Strack& Deutsch, 2004) suggest that such reductions
are likely to be highly contextual and short-lived. According to these
theories, implicit and explicit processes are supported by fundamental-
ly different psychological systems. Although the explicit system can
change quickly and is relatively context-independent, the implicit sys-
tem is highly contextual and only changes in an enduring way after con-
siderable time, effort, and / or intensity of experience. Thus, because
one-shot interventions must counteract a large accretion of associative
learning, they are unlikely to produce enduring change in the implicit
system. Such change is likely only after the application of considerable
goal-directed effort over time.
The preceding analysis is consistent with Devine's habit-breaking
analysis of prejudice reduction, which argues that overcoming preju-
dice is a protracted process that requires considerable effort in the
pursuit of a nonprejudiced goal (Devine, 1989; Devine & Monteith,
1993; Devine, Monteith, Zuwerink, & Elliot, 1991; Monteith, 1993).
This model likens implicit biases to deeply entrenched habits devel-
oped through socialization experiences. Breaking the habitof im-
plicit bias therefore requires learning about the contexts that
activate the bias and how to replace the biased responses with re-
sponses that reect one's nonprejudiced goals.
Supporting the prejudice habit-breaking framework, considerable
evidence demonstrates that, when they believe they have acted with
bias, people who endorse values opposed to prejudice are motivated
to inhibit the expression of implicit bias by seeking out information
and putting effort into tasks they believe would help them break the
prejudice habit (Amodio, Devine, & Harmon-Jones, 2007; Monteith,
1993; Plant & Devine, 2009). In addition, when these people act with
prejudice, they experience guilt (Devine et al., 1991), which instigates
self-regulatory efforts to disrupt automatic bias and prevent future ex-
pressions of bias (Amodio et al., 2007; Monteith, 1993). Although this
evidence is consistent with the prejudice habit-breaking framework,
extant research has not yet examined whether interventions can pro-
duce long-term implicit bias reductions, nor has it clearly specied the
type of effort required to yield such reductions. The goal of the present
work is to address these shortcomings and to develop an intervention
that engages intentional effort to produce enduring reductions in im-
plicit race bias.
Multifaceted prejudice habit-breaking intervention
Devine and colleagues (Devine & Monteith, 1993; Plant & Devine,
2009) argue that the motivation to break the prejudice habit stems
from two sources. First, people must be aware of their biases and, sec-
ond, they must be concerned about the consequences of their biases
before they will be motivated to exert effort to eliminate them. Fur-
thermore, people need to know when biased responses are likely to
occur and how to replace those biased responses with responses
more consistent with their goals.
The present work synthesizes insights from the prejudice habit
model and implicit bias reduction strategies to develop an intervention
to help people reduce implicit biases and break the prejudice habit.
The multifaceted nature of the present intervention has conceptual par-
allels to approaches in several other areas, such as health behavior
change (Prochaska & Velicer, 1997), cognitive behavior therapy (Beck
& Alford, 2009; Cox, Abramson, Devine, & Hollon, 2012), and the
fundamentals of adult learning (Howell, 1982; Kaufman, 2003). We
tested this intervention in a three-month longitudinal study, comparing
a group of people who completed the intervention to a control group
who did not.
To ensure situational awareness of their bias, all participants com-
pleted a measure of implicit bias and received feedback about their
level of bias. People assigned to the intervention group were also
presented with a bias education and training program, the goals of
which were to evoke a general concern about implicit biases and train
people to eliminate such biases. The education component likened the
expression of implicit biases to a habit and provided information linking
implicit bias to discriminatory behaviors acrossa wide range of settings
(e.g., interpersonal, employment, health). The training component de-
scribed how to apply a variety of bias reduction strategies in daily life.
Because the goal of our intervention was to engage a general
self-regulatory process, we did not present the strategies in separate
conditions to test each strategy's relative effectiveness. Instead, the
training section presented participants with a wide array of strategies,
enabling participants to exibly choose the strategies most applicable
to different situations in their lives. As part of the intervention, partici-
pants were prompted to report and reect on their strategy use in the
weeks between implicit bias assessments. We predicted that only peo-
ple who received the intervention would translate their situational
awareness into chronic awareness of biases in themselves and in socie-
ty, thereby ipping the self-regulatory switch that motivates strategy
use and reduces implicit bias.
To evaluate the effectiveness of the intervention, we examined its
impact on an indicator of implicit bias and a variety of explicit mea-
sures longitudinally. We used the BlackWhite Implicit Association
Test (IAT; Greenwald, McGhee, & Schwartz, 1998) as our measure of
implicit race bias. The explicit measures included established mea-
sures of racial attitudes (Brigham, 1993), the sources of one's motiva-
tion for responding without prejudice (Plant & Devine, 1998), and
whether one believes one's own behavior is more biased than appro-
priate (Monteith & Voils, 1998). Because 90% of our sample had a
pro-White bias on the BlackWhite IAT, the latter served as a measure
of awareness of one's tendency to respond with prejudice. In addition,
because the intervention included education about the adverse ef-
fects of discrimination, we developed a measure assessing concern
about discrimination in society. For both the intervention and control
groups, all measures were assessed prior to the intervention manipu-
lation and at two time points after the manipulation. We also asked
the intervention group participants a variety of questions immediate-
ly after the education and training program about the strategies they
had learned, and, in the weeks following the administration of the in-
tervention, we asked them some open-ended questions about their
use of the strategies.
Our design has ve major strengths. First, it allows us to assess the
intervention's effects on a rich array of variables (implicit and explic-
it) that are theoretically important to the reduction of race bias. Sec-
ond, it enables us to examine whether the intervention's effects on
these variables persisted or changed over time. Third, we have an op-
portunity to evaluate whether reported strategy use is associated
with reductions in implicit bias. Fourth, in the control group, we can
assess whether feedback about one's level of implicit bias leads to re-
ductions in implicit bias without a multifaceted intervention. Finally,
we can examine whether any of the explicit measurements taken at
two times, prior to and after the intervention manipulation, moderate
the effect of the intervention on implicit bias. A moderation effect
with a measure taken prior to the intervention would suggest that
the construct is related to learning processes during the intervention,
while a moderation effect with a measure taken after the intervention
would suggest that the construct is involved in the deployment of the
bias-reducing strategies. Together, these two sets of moderation anal-
yses can yield insight into two different aspects of the bias reduction
process.
1268 P.G. Devine et al. / Journal of Experimental Social Psychology 48 (2012) 12671278
Method
Participants and design
The participants were 91 non-Black introductory psychology stu-
dents (67% female, 85% White),
2
who completed a 12-week longitudi-
nal study for course credit (see Fig. 1). Attrition rates were low, never
exceeding 10% at any time point. Participants were randomly assigned
to either a control condition (n=38) or an intervention condition
(n=53), with more people assigned to the intervention condition to
provide greater power for analyses using the strategies measures and
because we anticipated greater attrition in the intervention condition
(but, attrition rates did not vary across condition at any time point, all
ps> .23). Throughout the study, participants completed the Black
White IAT and several explicit measures, described below. The IAT
was administered in the lab at three time points: just prior to the inter-
vention manipulation (baseline) and 4 and 8 weeks after the manipula-
tion. The explicit measures were also administered at three time points:
4 weeks prior to the manipulation in a classroom setting (baseline) and
2 and 6 weeks after the manipulation viaemail. The intervention group
participants also completed measures of their reactionsto the strategies
(e.g., perceived likelihood of use, perceived opportunity to use) imme-
diately following the intervention. Finally, intervention group partici-
pants completed free-response questions about their experiences
using the strategies 2 and 6 weeks after the intervention.
Implicit measure
Implicit race bias was measured with the BlackWhite IAT. TheIAT is
a dual-categorization task that has good psychometric properties
(Cunningham, Preacher, & Banaji, 2001; Hofmann et al., 2005)andis
linked to basic neural and affective processes relevant to implicit race
bias (Cunningham, Raye, & Johnson, 2004; Phelps et al., 2000). Addi-
tionally, in intergroup contexts, the IAT is a strong predictor of discrim-
inatory behavior and a better predictor than parallel explicit measures
(Greenwald, Poehlman, Uhlmann, & Banaji, 2009; but see Blanton et
al., 2009 for an opposing view).
In the BlackWhite IAT, people categorize sequentially presented
stimuli based on whether they are pleasant or unpleasant words or
Black or White faces bypressing keys on the left or right side of the key-
board. Underlying the race IAT is the assumption that, to the extent that
negative valence is associated with Black people (and positive valence
with White people), when Black faces and unpleasant words (and
White faces and pleasant words) are paired together on the same re-
sponse key (compatible trials), the task should be easier than with the
reverse pairings (incompatible trials). Response times on the compatible
and incompatible trials are used to compute D-scores (see Greenwald,
Nosek, & Banaji, 2003). Higher D-scores indicate that participants
more easily associate Black faces with unpleasant words (and White
faces with pleasant words) than the reverse (baseline M=.46, SD =
.39, skew= .16, split-half reliability=.60).
Explicit measures
We assessed a variety of explicit measures that have been implicat-
ed in the bias-reducing process, including racial attitudes, the sources of
motivation to respond without prejudice, prejudice-relevant discrepan-
cies, and concern about discrimination in society.
Racial attitudes
Racial attitudes were assessed using the Attitudes Towards Blacks
scale (ATB; Brigham, 1993). The ATB has 20 items, each assessed on a
1(strongly disagree)to7(strongly agree) scale. Responses to the items
are averaged together, and higher scores indicate more positive atti-
tudes towards Blacks and therefore, less explicit race bias (baseline
M=7.57, SD=.68, skew=.53, α=.82).
Motivations to respond without prejudice
Plant and Devine (1998) distinguish between internal motivation
to respond without prejudice, which is primarily driven by personal
values and the belief that prejudice is wrong, and external motivation
to respond without prejudice, which is primarily driven by a desire to
escape social sanctions. We used the Internal Motivation Scale and Ex-
ternal Motivation Scale (IMS and EMS; Plant & Devine, 1998)toassess
these separate motivations. Both the IMS and EMS are composed of 5
items, each assessed on a 1 (strongly disagree)to9(strongly agree)
scale. For each scale, participants' responses on the items are averaged
together such that higher scores indicate more motivation to respond
without prejudice (baseline IMS M=7.03, SD= 1.17, skew= .95,
α=.72, EMS M=5.43, SD = 1.78, skew= .01, α=.80).
Shoulds, woulds, and discrepancies
The discrepancy scale measures the extent to which people predict
they would act with more prejudice than they believe is appropriate. Be-
cause most people do have some implicit bias, it can also be used as an in-
dicator of one's personal awareness of one's bias (Monteith & Voils, 1998).
The full scale is composed of should and would subscales, whose items use
a1(strongly disagree)to7(strongly agree)scale.Theshould subscale asks
people how they believe they should act, feel, or think in response to nine
interpersonal intergroup situations (e.g., I should feel uncomfortable sit-
ting next to a Black person on a bus). Responses to the items are averaged
together such that higher numbers indicate a standard that permits more
prejudice (baseline M=1.98, SD =.68, skew = 1.68, α=.62). The would
subscale asks people to predict how they would actually act, feel, or
think in each of the situations (e.g., I would feel uncomfortable sitting
next to a Black person on a bus). Responses to the items are averaged to-
gether such that higher numbers indicate predictions that the person
would act with greater prejudice (baseline M=3.18, SD=.95, skew=
.10, α=.77). Prejudice-relevant discrepancies are calculated by rst
subtracting the score for each should from its corresponding would,then
averaging the resulting differences. For this computed score, more posi-
tive numbers indicate a greater belief that one would act with more prej-
udice than one believes is appropriate in intergroup situations (baseline
M=1.20, SD = .86, skew = .17, α=.65).
Concern about discrimination
One of the goals of the intervention was to educate people about the
existence of discrimination. We therefore developed a measure of be-
liefs that discrimination is a problem in society. This measure comprises
four items, each measured on a 1 (strongly disagree)to10(strongly
agree) scale. Item responses were averaged, resulting in a score for
which higher numbers indicate greater concern (see Appendix A for
the full measure; baseline M=6.08, SD= 1.22, skew= .15, α=.86).
Strategy measures
During the rst lab session, following the intervention manipulation,
the intervention group completed a set of Likert-type questions designed
to assess the participants' reactions to each of the individual strategies.
These questions used a 1 (not at all)to7(very much) scale and included
items assessing the perceived likelihood of using each strategy, the will-
ingness to use each strategy, the perceived difculty of implementing
each strategy, the perceived effectiveness of each strategy, and the per-
ceived opportunities to use each strategy. The participants' responses
to each item for all six techniques were averaged to obtain mean likeli-
hood (M=4.56, SD =1.09, skew=.21, α=.87), willingness (M=
2
We recruited non-Black participants of all races because of evidence that non-Black
participants of all races show similar levels of implicit anti-Black bias (Nosek, Banaji, &
Greenwald, 2002). On the basis of this evidence, we also included the non-Black partic-
ipants in all analyses.
1269P.G. Devine et al. / Journal of Experimental Social Psychology 48 (2012) 12671278
5.78, SD= .86, skew = .20, α= .86), difculty (M=3.74, SD =1.27,
skew=.28, α=.85), effectiveness (M= 4.88, SD =.72, skew .03, α=
.68), and opportunity scores (M=4.34, SD = 1.19, skew = .15, α=
.83), on which higher numbers indicate higher likelihood, willingness,
perceived difculty, perceived effectiveness, and perceived opportuni-
ties, respectively.
Strategy use free-response questions
At the 2 and 6 week explicit measure assessments, the intervention
group also completed questionnaires in which they gave open-ended
responses about their experiences using the strategies. For each strate-
gy, the participants were asked whether they had used the strategy
since their last in-lab session. If they had used a strategy, the partici-
pants were subsequently asked to describe one or two situations in
which they had used the strategy and to provide general comments
about their experiences using the strategy. At the end of the question-
naire, the participants were asked to share any additional comments
about implementing the strategies.
Procedure
During the rst lab session, all participants completed the IAT and
received feedback abouttheir performance. Specically, an experiment-
er calculated the participants' IAT scores and asked them to type their
scores into the computer. The computer provided participants with an
interpretation of their IAT performance based on their D-scores, saying
that they had a strong preference for Blacks over Whites (D-score less
than or equal to .65), a moderate preference for Blacks over Whites
(D-score between .65 and .35), a slight preference for Blacks over
Whites (D-score between .35 and .15) no preference for Whites
or Blacks (D-score between .15 and .15), a slight preference for
Whites over Blacks (D-score between .15 and .35), a moderate prefer-
ence for Whites over Blacks (D-score between .35 and .65) or a strong
preference for Whites over Blacks (D-score over .65).
To the extent that the participants implicitly favored White people
over Black people (as 90% of our participants did), we expected that
being confronted with evidence of this bias would increase participants'
awareness of their bias (Monteith, Voils, & Ashburn-Nardo, 2001). After
receiving feedback, control group participants were dismissed; they
were reminded, however, that they would return to the lab at two sub-
sequent points in time and would receive questionnaires to ll out be-
tween their lab sessions. People in the intervention group were
presented with a 45-minute narrated and interactive slideshow sepa-
rated into education and training sections.
3
The education section intro-
duced the idea of prejudice as a habit, as well as how implicit biases
develop and are automatically activated without intention. This section
also explained the general logic behind how the IAT measures implicit
bias (without giving the participants a specic strategy to beat the
IAT;Kim, 2003) and dispelled alternate explanations for IAT bias
(e.g., the order of the congruent and incongruent trials, color associa-
tions). Participants were then taught about the prevalence of implicit
race biasesand how they can lead people to unwittingly perpetuate dis-
crimination. Specically, participants learned about research linkingIAT
bias to a wide range of discriminatory outcomes in domains such as
health, employment, and everyday interpersonal interactions.
Strategies for reducing implicit race bias
The training section provided participants with a list of ve strategies
culled from the literature and adapted for the intervention (see Fig. 1).
The program explained the strategies in straightforward language with
concrete examples of everyday situations in which they could be used.
Participants were then asked to generate situations in which they could
use each strategy. Participants were told that although none of the strat-
egies are difcult to implement, each requires some effort. In addition, the
program emphasized how the strategies (explained below) are mutually
reinforcing. For example, contact with counter-stereotypic others pro-
vides grist for counter-stereotypic imaging as well as providing opportu-
nities for individuation, perspective taking and stereotype replacement.
Similarly, perspective taking can enhance stereotype replacement and in-
dividuation by encouraging people to see the world from the eyes of a
stigmatized other. As a set, the strategies were offered as a powerful
toolkit for breaking the prejudice habit. The program also stressed that
practicing the strategies would help them to reduce implicit bias and,
hence, break the prejudice habit. Following the education and training
sessions, participants were reminded that they would return to the lab
for two subsequent sessions and would receive questionnaires to com-
plete between the lab sessions. Participants were then dismissed.
Stereotype replacement
This strategy involves replacing stereotypical responses for non-
stereotypical responses. Using this strategy to address personal
stereotyping involves recognizing that a response is based on stereo-
types, labeling the response as stereotypical, and reecting on why
the response occurred. Next one considers how the biased response
could be avoided in the future and replaces it with an unbiased re-
sponse (Monteith, 1993). A parallel process can be applied to societal
(e.g., media) stereotyping.
Counter-stereotypic imaging
This strategy involves imaginingin detail counter-stereotypic others
(Blair et al., 2001). These others can be abstract (e.g., smart Black
3
Materials available upon request to the rst author.
Fig. 1. Study timeline. The BlackWhite Implicit Association Test (IAT) was administered at 3 time points: just prior to the intervention manipulation and 4 and 8 weeks after the
manipulation. The explicit measures, consisting of the Attitudes Towards Blacks (ATB) scale, the Internal and External Motivation Scales (IMS and EMS), the prejudice-relevant dis-
crepancies scale, and the concern about discrimination scale, were also administered at three points: 4 weeks prior to the intervention manipulation during a mass survey, and 2
and 6 weeks after the manipulation.
1270 P.G. Devine et al. / Journal of Experimental Social Psychology 48 (2012) 12671278
people), famous (e.g., Barack Obama), or non-famous (e.g., a personal
friend). The strategy makes positive exemplars salient and accessible
when challenging a stereotype's validity.
Individuation
This strategy relies on preventing stereotypic inferences by obtaining
specic information about group members (Brewer, 1988; Fiske &
Neuberg, 1990). Using this strategy helps people evaluate members of
the target group based on personal, rather than group-based, attributes.
Perspective taking
This strategy involves taking the perspective in the rst person of
a member of a stereotyped group. Perspective taking increases psy-
chological closeness to the stigmatized group, which ameliorates au-
tomatic group-based evaluations (Galinsky & Moskowitz, 2000).
Increasing opportunities for contact
This strategy involves seeking opportunities to encounter and engage
in positive interactions with out-group members. Increased contact can
ameliorate implicit bias through a wide variety of mechanisms, including
altering the cognitive representations of the group or by directly improv-
ing evaluations of the group (Pettigrew, 1998; Pettigrew & Tropp, 2006).
Results
Data analytic plan
The intervention and control groups did not differ on any of the
measures at baseline, all ps.50 (see Table 1; for correlations between
the study variables within each condition, see Table 2). All of the analy-
ses were conducted using a series of General Linear Models (GLMs)
using the baseline measurement of the dependent variable as a covari-
ate, an approach that ismore powerful than using a difference score ap-
proach when there are no pre-test differences on the dependent
variable (Van Breukelen, 2006). Predictors were centered prior to test-
ing interactions. All missing data were handled through multiple impu-
tation (Rubin, 1987).
4
To examine the multifarious effects of the intervention after base-
line, we conducted a series of analyses. First, we analyzed the effect of
the intervention on ourimplicit measure of race bias. We then analyzed
the effect of the intervention on each of our explicit measures. Next, we
conducteda series of moderation analyses of the impact of the interven-
tion on implicit bias using the baseline and week 2 measurements of
each of the explicit measures. Finally, we conducted a seriesof analyses
within the intervention group to determine whether either the
post-manipulation reactions to the strategies or the coded variables
from the free response reports of strategy use predicted reductions in
implicit bias.
Effect of the intervention on implicit race bias
A major goal of this study was to examine the impact of the inter-
vention on the magnitude of implicit race bias assessed over time. As
shown in Fig. 2, the intervention was successful. Following the manipu-
lation, intervention group participants had lower IAT scores than con-
trol group participants, B=.19, t(88)= 2.82, p=.006, ΔR
2
=.081.
Moreover, the effects of the intervention on implicit race bias at 4 and
8 weeks were not systematically different from each other, B=.091,
t(88)= .82, p=.42, ΔR
2
=.008, indicating that the reduction in implicit
race bias persisted throughout the 8-week interval. These data provide
the rst evidence that a controlled, randomized intervention can pro-
duce enduring reductions in implicit bias.
5
Effect of the intervention on the explicit measures
We were also interested in whether the intervention affected explicit
variables previously argued to be important to the prejudice reduction
process. The intervention manipulation created no changes in either the
participants' reported racial attitudes or their internal/external motiva-
tions to respond without prejudice (all ps.53).Itdid,however,affect
participants' concern about discrimination and their awareness of their
personal bias as revealed through their prejudice-relevant discrepancies.
Intervention group participants had higher concern than control group
participants following the intervention manipulation, B= .43, t(88)=
2.24, p=.028, ΔR
2
=.049. As shown in Fig. 3, participants in the control
group had a constant level of concern throughout the duration of the
study, B=.044, t(89)= .22, p=.83, d=.030, whereas participants
in the intervention group increased in concern after they received the in-
tervention, B= .40, t(89) = 2.10, p= .042, d= .38. The effect of the inter-
vention manipulation on concern also grew more pronounced over time,
B=.54, t(88)= 2.35, p= .021, ΔR
2
=.061. We explored whether the in-
creases in concern observed in the intervention group predicted the de-
creases in implicit race bias, but this was not the case, B=.033,
t(88)=.60, p=.55, ΔR
2
=.005, suggesting that the decreases in im-
plicit bias and increases in concern were somewhat independent effects
of the intervention.
Table 1
Means and standard deviations of implicit and explicit variable by condition.
Intervention
condition (N=53)
Control condition
(N=38)
Mean SD Mean SD
Baseline IAT D-score 0.47 0.42 0.42 0.36
ATB 5.67 0.66 5.57 0.72
EMS 5.37 1.80 5.52 1.75
IMS 7.64 0.97 7.47 1.41
Shoulds 2.03 0.65 1.91 0.72
Woulds 3.16 1.00 3.20 0.88
Discrepancies 1.17 0.90 1.29 0.80
Concern 6.08 1.06 6.05 1.37
Time 2 IAT D-score 0.32 0.41 0.54 0.36
ATB 5.52 0.71 5.58 0.72
EMS 5.02 1.87 5.52 1.75
IMS 7.11 1.19 7.47 1.41
Shoulds 1.90 0.55 1.99 0.78
Woulds 3.51 1.04 3.29 0.94
Discrepancies 1.63 1.07 1.34 1.06
Concern 6.42 0.95 6.24 1.20
Time 3 IAT D-score 0.30 0.42 0.47 0.41
ATB 5.60 0.73 5.53 0.72
EMS 5.00 1.64 5.21 1.95
IMS 7.28 1.28 7.28 1.43
Shoulds 1.89 0.54 1.90 0.67
Woulds 3.38 0.93 3.15 0.85
Discrepancies 1.49 0.86 1.26 0.76
Concern 6.57 1.04 5.86 1.24
Note: None of the measures differ by condition at baseline.
4
We used ve imputations, each of which were iterated until convergence. All re-
sults were pooled Rubin's rules. Although we report unadjusted degrees of freedom,
all other reported statistics are adjusted for the amount of information lost due to
missingness.
5
An anonymous reviewer brought up the possibility that the participants in the
training condition spontaneously discovered strategies to beat the IATafter they re-
ceived information on how the IAT measures implicit bias (Kim, 2003). To guard
against this possibility, the reviewer suggested analyzing response latencies on the
congruent and incongruent trials separately to determine whether the participants
were deliberately slowing their responses on the congruent trials. After the interven-
tion was administered, participants in the training and control groups had equal laten-
cies on both congruent trials, B=18.75, t(88)= .76, p= .45, ΔR
2
=.006, and on
incongruent trials, B=34.24, t(88)=1.02, p= .31, ΔR
2
=.011.
1271P.G. Devine et al. / Journal of Experimental Social Psychology 48 (2012) 12671278
Table 2
Correlations between the major study variables within the control and training conditions. Correlations in the control condition are shown below the diagonal, whereas correlations within the training condition are shown above the
diagonal.
IAT 1 IAT 2 IAT 3 ATB 1 ATB 2 ATB 3 EMS 1 EMS 2 EMS 3 IMS 1 IMS 2 IMS 3 Should
1
Should
2
Should
3
Would
1
Would
2
Would
3
Discrepancies
1
Discrepancies
2
Discrepancies
3
Concern
1
Concern
2
Concern
3
IAT 1 0.36 0.01 0.14 0.20 0.15 0.12 0.16 0.13 0.05 0.08 0.16 0.09 0.08 0.05 0.15 0.25 0.24 0.11 0.20 0.22 0.07 0.08 0.11
IAT 2 0.33 0.27 0.27 0.38 0.35 0.21 0.31 0.25 0.22 0.23 0.19 0.26 0.24 0.24 0.33 0.31 0.31 0.19 0.19 0.18 0.28 0.36 0.10
IAT 3 0.21 0.18 0.01 0.09 0.05 0.03 0.14 0.09 0.09 0.01 0.11 0.13 0.11 0.10 0.03 0.00 0.10 0.06 0.05 0.04 0.03 0.34 0.1
1ATB 1 0.15 0.17 0.02 0.74 0.75 0.16 0.03 0.14 0.37 0.48 0.44 0.65 0.58 0.66 0.58 0.43 0.44 0.19 0.13 0.04 0.42 0.30 0.42
ATB 2 0.20 0.19 0.01 0.77 0.83 0.22 0.05 0.12 0.34 0.63 0.50 0.46 0.57 0.66 0.54 0.59 0.55 0.29 0.30 0.16 0.31 0.42 0.48
ATB 3 0.08 0.19 0.00 0.73 0.68 0.24 0.15 0.27 0.35 0.60 0.48 0.51 0.64 0.71 0.62 0.57 0.64 0.35 0.24 0.22 0.40 0.46 0.53
EMS 1 0.18 0.01 0.16 0.04 0.03 0.10 0.43 0.56 0.01 0.11 0.25 0.15 0.15 0.05 0.37 0.19 0.14 0.33 0.11 0.11 0.18 0.16 0.15
EMS 2 0.01 0.04 0.02 0.06 0.07 0.05 0.46 0.66 0.07 0.24 0.01 0.09 0.05 0.06 0.19 0.39 0.31 0.16 0.40 0.38 0.23 0.21 0.03
EMS 3 0.08 0.09 0.02 0.05 0.08 0.11 0.37 0.67 0.14 0.14 0.08 0.15 0.02 0.06 0.36 0.37 0.25 0.31 0.35 0.31 0.31 0.40 0.17
IMS 1 0.04 0.09 0.16 0.64 0.69 0.32 0.04 0.10 0.10 0.31 0.44- 0.41 0.40 0.23 0.39 0.36 41 0.15 0.16 0.30 0.28 0.11 0.12
IMS 2 0.04 0.20 0.12 0.39 0.58 0.30 0.11 0.30 0.08 0.75 0.77 0.44 0.63 0.64 0.32 0.24 0.32 0.04 0.07 0.08 0.16 0.33 0.42
IMS 3 0.19 0.27 0.04 0.58 0.81 0.48 0.00 0.24 0.00 0.77 0.80 0.49 0.60 0.63 0.28 0.22 0.28 0.05 0.08 0.11 0.25 0.16 0.37
Should 1 0.23 0.26 0.05 0.51 0.49 0.55 0.11 0.25 0.08 0.40 0.46 0.51 0.66 0.58 0.57 0.35 0.40 0.10 0.02 0.05 0.30 0.19 0.23
Should 2 0.18 0.37 0.08 0.48 0.56 0.55 0.22 0.35 0.21 0.44 0.58 0.64 0.70 0.75 0.47 0.20 0.32 0.05 0.29 0.15 0.20 0.26 0.36
Should 3 0.38 0.38 0.16 0.58 0.71 0.61 0.14 0.38 0.20 0.38 0.49 0.65 0.66 0.73 0.32 0.37 0.44 0.07 0.01 0.19 0.31 0.26 0.48
Would 1 0.05 0.15 0.08 0.61 0.50 0.61 0.07 0.01 0.19 0.26 0.35 0.39 0.38 0.33 0.48 0.62 0.65 0.76 0.37 0.50 0.22 0.08 0.16
Would 2 0.13 0.19 0.12 0.49 0.64 0.40 0.11 0.28 0.21 0.48 0.35 0.49 0.19 0.21 0.32 0.51 0.85 0.47 0.88 0.69 0.25 0.25 0.20
Would 3 0.05 0.31 0.01 0.69 0.74 0.64 0.17 0.11 0.16 0.57 0.40 0.66 0.34 0.41 0.56 0.62 0.76 0.48 0.67 0.80 0.18 0.03 0.12
Discrepancies
1
0.14 0.06 0.11 0.19 0.09 0.16 0.16 0.19 0.25 0.07 0.03 0.03 0.43 0.24 0.06 0.67 0.35 0.34 0.44 0.57 0.03 0.05 0.01
Discrepancies
2
0.01 0.10 0.16 0.08 0.15 0.04 0.25 0.49 0.33 0.10 0.11 0.03 0.33 0.53 0.24 0.21 0.72 0.37 0.47 0.74 0.15 0.12 0.02
Discrepancies
3
0.29 0.01 0.13 0.26 0.21 0.18 0.33 0.48 0.37 0.31 0.02 0.16 0.21 0.19 0.27 0.28 0.59 0.65 0.44 0.65 0.01 0.14 0.19
Concern 1 0.15 0.18 0.05 0.53 0.48 0.26 0.07 0.37 0.23 0.54 0.46 0.46 0.12 .21 0.31 0.45 0.25 0.34 0.34 0.07 0.10 0.39 0.27
Concern 2 0.00 0.10 0.07 0.23 0.56 0.32 0.05 0.25 0.03 0.52 0.55 0.60 0.31 0.26 0.42 0.17 0.27 0.36 0.07 0.05 0.03 0.41 0.53
Concern 3 0.07 0.14 0.06 0.33 0.53 0.49 0.04 0.21 0.09 0.36 0.31 0.57 0.27 0.34 0.41 0.17 0.23 0.41 0.05 0.04 0.10 0.29 0.66
1272 P.G. Devine et al. / Journal of Experimental Social Psychology 48 (2012) 12671278
Compared to control group participants, intervention group par-
ticipants reported greater discrepancies between their shouldsand
woulds following the intervention, B= .38, t(88) = 2.31, p=.024,
ΔR
2
=.047. Separate analyses on the should and would indices re-
vealed that although the standards for responding towards Black
people (shoulds) remained unchanged in the intervention group rel-
ative to the control group, B=.095, t(88) =.93, p= .36, ΔR
2
=
.005, participants in the intervention group increased in how much
they predicted they would respond with bias in intergroup situatio ns
(woulds) following the intervention manipulation, B= .29, t(88)=
1.93, p=.057, ΔR
2
=.026. As a set, the analyses on shoulds, woulds,
and prejudice-relevant discrepancies suggest that the intervention
caused people to become more aware of their personal bias, while
leaving people's standards for behavior in prejudice-relevant situa-
tions unchanged. As with concern, we explored whether the in-
creases in discrepancies predicted the decreases in implicit race
bias, but this was not the case, B=.037, t(88) = .67, p=.50,
ΔR
2
=.005. The effect of the intervention on discrepancies was not
systematically different at week 4 and week 8, B=.067,
t(88)= .37, p= .72, ΔR
2
=.004 (see Fig. 4).
Moderation analyses
We next investigated whether any of the explicit measurements
taken at baseline or week 2 moderated the effect of the intervention
on implicit bias. Somewhat surprisingly, none of the explicit measures
taken at baseline moderated the effect of the intervention on implicit
bias, all ps.15.
6
The only week 2 explicit variable toemerge as a mod-
erator was concern about discrimination, B=.15, t(86)=2.46, p=
.016, ΔR
2
=.058, all other ps.23. As shown in Fig. 5, intervention con-
dition participants with more concern about discrimination at week 2
had particularly low levels of implicit bias at weeks 4 and 8. This effect
remained from week 4 to week 8, B=.043, t(86)= .082, p= .94,
ΔR
2
=.002, indicating that people high in concern about discrimination
at week 2 retained the reductions in IAT bias 8 weeks after the
intervention. The interactive effect of condition and concern was entire-
ly driven by a robust relationship between concern and implicit bias in
the intervention condition, B=.16, t(86)= 2.90, p=.009, ΔR
2
=
.12, a relationship that was entirely absent in the control condition,
B=.028, t(86)= .55, p=.59, ΔR
2
=.005.
We believe that these interactive effects of the intervention manip-
ulation and concern on implicit race bias are of large practical signi-
cance. Compared to high concern control group participants, the
predicted IAT scores of high concern participants in the intervention
Fig. 4. Discrepancies betweenself-reportedstandards (shoulds)and predicted actualreac-
tions (woulds) to Blacksby condition 4 weeks before the manipulation and 2 and 6 weeks
after the manipulation. Higher numbers indicate that participants predict they would
react toBlacks with more bias thanthey believe is appropriate. Discrepanciesdid not differ
by condition before the manipulation, but after the manipulation, participants who re-
ceived the intervention had larger discrepancies than participants who did not. Error
bars represent± 1 standard error of the GLM point estimate.
6
The three-way interaction between internal motivation, external motivation, and
condition was also not signicant, B=.028, t(82) =.67, p=.52, ΔR
2
=.023.
Fig. 2. IAT D-scores for intervention and control group participants before the manipula-
tion and 4 and 8 weeks after the manipulation. Higher numbers indicate higher levels of
implicit bias.IAT D-scores did not differ before the manipulation, but after the manipula-
tion, participants who received the intervention had lower IAT scores than participants
who did not. Error bars represent ±1 standard error of the GLM point estimate.
Fig. 3. Concern about discrimination by condition 4 weeks before the manipulation and2
and 6 weeks after the manipulation. Higher numbers indicate higher levels of concern.
Concern did not differ by condition before the manipulation, but after the manipulation,
participants who received the intervention were more concerned about discrimination
than participants who did not. Error bars represent± 1 standard error of the GLM point
estimate.
1273P.G. Devine et al. / Journal of Experimental Social Psychology 48 (2012) 12671278
group were .38 units lower at week 4 and .31 units lower at week 8.
These decreases are, for example, large enough to bring someone from
the moderate preference for Whites over Blacksfeedback category
(D-score between .35 and .65) down into the slight preference for
Whites over Blacksfeedback category (D-score between .15 and .35).
The interaction remained signicant when controlling for all the other
explicit measures, B=.14, t(82)=2.31, p=.024, ΔR
2
=.049.
Perceptions of strategies analyses
To gain some understanding about what the participants in the in-
tervention condition did to reduce their implicit race bias, we examined
whether, after controlling for baseline implicit bias, the participants' re-
actions to the strategies were associated with lower IAT bias at week 4
and week 8. These analyses revealed that self-reported likelihood to use
the strategies was associated with lower implicit race bias, B=.14,
t(50)= 3.18, p=.003, ΔR
2
= .17, an effect that was notsystematically
different between week 4 and week 8, B=.050, t(50)= .69, p=
.50, ΔR
2
=.014. Though the effects were weaker, perceived opportuni-
ties also emerged as a predictor of lower implicit race bias, B=.092,
t(50)= 2.40, p=.021, ΔR
2
=.10, as did perceived effectiveness,
B=.12, t(50)= 1.86, p=.069, ΔR
2
=.063. Perceived opportunities
was, however, highly correlated with likelihood, r=.76, as was effec-
tiveness, r=.40, and when all three variables were allowed to predict
post-baseline implicit race bias, only the relationship with likelihood
remained robust, B=.13, t(48)=2.03, p=.051, ΔR
2
=.078. Thus,
to have a reduction in implicit race bias, it appears that participants
had not only to perceive opportunities to implement the strategies
and view the strategies as effective, but also to believe that they were
likely to use them. Perceived difculty of implementing the strategies
and willingness to use the strategies did not emerge as predictors of re-
duced bias, ps>.23.
Strategy use free-response analyses
Participants' descriptions of their strategy use were rich and com-
plex. To capture this complexity while maintaining objectivity, we
used a text mining approach to calculate frequencies of theoretically
important word stems in the descriptions. The overarching goal of this
approach was to determine if the constructs thought to be important
in the habit-breaking model were included in participants' descriptions,
and, if so, if they were also related to reductions in implicit bias.
To that end, we used the tm package for R(Feinerer, Hornik, &
Meyer, 2008) to load the participants' responses from each time point
into a computerized corpus of responses. Each response was screened
for the default stop words from the tm package (e.g., the,a,
and), and the resulting sets of words were reduced to word stems.
We then chose a standard psycholinguistic dictionary (WordNet;
Fellbaum, 1998) to look up synonyms of words that related to the
three theoretically important categories in the prejudice habit model:
motivation (or the decision that prejudice is wrong), awareness, and
the implementation of strategies to combat bias. After validating that
the meanings of all the resulting words matched that of the target cat-
egories, we eliminated all words that had not been used by at least
two people at each time point. The resulting word stems for each cate-
gory, as well as the means and standard deviations of their frequency of
use, are displayed in Table 3. Sample participant responses with bolded
target words are displayed in Table 4.
Fig. 5. Week 2 concern about discrimination plotted against week 4 (Panel A) and week
8 (Panel B) IAT D-scores with prediction lines from the GLM. Higher numbers indicate
higher levels of implicit bias and greater levels of concern. Prediction lines are plotted
at ±1standard deviation from the mean on concern. Within the interv ention condition,
concern was associated with lower IAT scores at weeks 4 and 8. Within the control con-
dition, concern was unrelated to IAT scores.
Table 3
Means and standard deviations of word stem frequencies from the free-response an-
swers concerning strategy use, broken into three different conceptual categories.
Week 2 Week 6
Mean SD Mean SD
Awareness stems awar 0.15 0.50 0.09 0.35
realiz 0.92 1.02 0.81 0.88
recogn 0.25 0.59 0.11 0.32
understand 0.17 0.58 0.09 0.30
Total 1.49 1.30 1.11 1.12
Motivation stems wrong 0.26 0.56 0.15 0.36
unfair 0.38 0.66 0.09 0.40
Total 0.64 0.98 0.25 0.65
Implementation stems implement 0.49 0.89 0.87 1.49
practic 0.11 0.32 0.21 0.53
appli 0.26 0.65 0.30 0.72
use 0.45 0.93 0.91 1.29
tri 0.64 0.94 0.40 0.77
Total 1.96 2.09 2.68 2.67
1274 P.G. Devine et al. / Journal of Experimental Social Psychology 48 (2012) 12671278
Table 4
Sample participant responses to the free-response questions regarding strategy use and their corresponding word stem frequencies for each conceptual category. Word stems included in the analyses are highlighted in bold.
Category Frequency Response
Awareness 3 I was at a house party one weekend when two tall and strong African Americans walked in. I immediately assumed they must be on the football or some other sports team.
I realized that thought was stereotypical and decided that such an assumption should not be made. It is very useful in evaluating one’s thoughts. Two of my favorite shows
are CSI and Law and Order: SVU. In many instances I was able to implement this technique. I could pick out stereotypes and realize they were unfair. I think this
technique is very useful because it shows that discrimination is not something a person may just do personally, but the media discriminates as well. Two of my cousins are
African American, so it is easy to implement this technique, especially when racist jokes are made. It is very useful for people who have someone in their life that does not
prove the stereotypes true. At the party I described above I chose to talk to the guys and get to know about them. It ended up that both were indeed on the football team but
I found that there was so much more to them then that, that single fact no longer seemed important. Although not always possible, this technique is useful when it can be
used. I implemented this technique during the same party situation, I realized how unfair I would have been to be seen and appreciated for only my athletic abilities. I
think this technique is best used following a situation, as a post analysis. I don’t always remember or think about using the techniques.
Motivation 2
Implementation 8
Awareness 2 Before spring break I was riding the bus and an older black gentlemen sat in the seat next to me. I was about to sit closer to the wall but then realized that this was a
stereotypical response and stayed where I was. This technique is very effective and very easy to implement in daily activities. I had to go get a drug test for my new job.
While I was in the clinic a black woman walked up to the desk and the receptionist assumed she was there for a drug test also. I thought that was stereotyping on her part
and I was right. The woman was there because she was injured at work. This technique is easy to recognize in a society filled with stereotypes and prejudice. Over break I
was driving to work and a black woman flew by me on 151. I immediately figured she was rushing to work and was late all the time. Really I had no clue what the situation
was. This could be her first time being late to work, or maybe she just likes to speed. I didn’t know the individual so I was simply generalizing. This process requires a little
bit of thought but it is good to look at people as individuals and not lump people together. This process seems like it would be very difficult because I have never been
black and have no clue how hard it would be to grow up dealing with stereotypes and prejudice. I recently got a job at an employer that actually believes in equal
opportunity and I have many black co workers. I have decided to try to get to know my black coworkers better since at my last job they did not really hire black people and
I had no opportunity for this. This technique requires the most physical work but should be easy if stereotypes are put aside. 1, 2, 4 are quite natural to do. 3, 5, and 6 are
sometimes unnatural and take a little more work to implement. Although it’s not always easy using these techniques they are a good way to work at getting rid of
stereotypes we may have learned from personal experience or the media. It would have been nice if we would have been told to start implementing these techniques
immediately following session 1.
Motivation 0
Implementation 5
Awareness 3 The movie showed a black male with an afro, blowing things up and stealing things. I realize this is stereotypical and replace it. I don’t feel like this type of action in the
media creates much impact on a racial level. I was walking down an alley when a black man approached me and my first response was to stay as far away from him as
possible because I was afraid he would mug me seeing that I had cash in my hand. I realized this man was homeless and not scary at all but rather trying to get some food.
This is one of the easy techniques to apply because it is quite easy to establish that something is stereotypical and alter the way you think about the situation. I recently had
a party and a black male showed up, I was hesitant to let him wander around in my house because I didn’t know him that well but I had no issue with the same situation
white males. Instead I talked with him for nearly 30 minutes and realized I had nothing to worry about. This technique comes across fairly easy if the situation presents
itself to you. I recently went to a dance event sponsored by a group majority of black students. This is an easy technique to gain knowledge of a different race or religious
view. Increasing opportunities Individuating. It doesn’t seem that difficult but I don’t feel like it changes my subconscious attitudes towards blacks.
Motivation 0
Implementation 2
Awareness 1 Scanning IDs at the Nat, on spring break in myrtle beach I realized I was thinking stereotypical thoughts when I saw a black person walking down the street and was
afraid, so I remembered all the times this happened and nothing happened. Upon thinking a black person might not be as smart as others in the class I imagined what that
would feel like if people thought of me like that. I think it is a good way to prevent stereotypical thoughts.
Motivation 0
Implementation 0
1275P.G. Devine et al. / Journal of Experimental Social Psychology 48 (2012) 12671278
We then t a series of models to determine whether, after control-
ling for baseline implicit bias, word stem use from each conceptual cat-
egory at week 2 was related to implicit bias at week 4 or week 8 and
whether word stem use at week 6 related to implicit bias at week 8.
The only signicant effect to emerge from these analyses was that
more frequent use of implementation-related word stems at week 2
predicted reduced implicit bias at week 4, B=.068, t(50)= 2.47,
p=.017, ΔR
2
=.11, all other ps> .21. We also tested whether likelihood
was related to use of implementation-related word stems, but, in fact,
these two variables were almost entirely independent of each other,
r=.027. When used to simultaneously predict week 4 implicit bias,
both likelihood and implementation-related word use were related to
decreased implicit bias, B=.13, t(49)= 2.40, p=.028, ΔR
2
=.12,
and B=.068, t(49)= 2.66, p=.012, ΔR
2
=.11, respectively. Likeli-
hood and week 2 implementation-related word stem use jointly
accounted for fully 23% of the variance in week 4 implicit bias.
Discussion
Overall, our results provide compelling and encouraging evidence
for the effectiveness of our multifaceted intervention in promoting
enduring reductions in implicit bias. As such, this study provides a re-
sounding response to the clarion call for methods to reduce implicit
bias and thereby reduce the pernicious, unintended discrimination
that arises from implicit biases. Reductions in implicit bias that emerged
by week 4 following the intervention persisted to week 8. Such enduring
reductions in implicit bias following a bias reduction intervention are un-
precedented in the literature (Paluck & Green, 2009). Although some pre-
vious research has established that people who choose to immerse
themselves in a context either rich in counter-stereotypic exemplars
(e.g., a women's college; Dasgupta & Asgari, 2004)orthatisconducive
to the regular discussion of issues related to implicit bias (e.g., in a course
on stereotyping and prejudice; Rudman,Ashmore,&Gary,2001)show
reduced implicit bias, our study is the rsttoourknowledgetoproduce
long-term change in implicit bias using a randomized, controlled design.
Another encouraging nding was the simultaneous effect of the in-
tervention on increasing people's self-reported concern about discrimi-
nation and prejudice-relevant discrepancies. The intervention thus
seems to increase both personal awareness of one's bias and a general
concern about discrimination in society. The effect of the intervention
on concern also grew more pronounced over time, potentially suggesting
that the intervention created an increased caring about subtle instances
of bias and discrimination. We suspect that the intervention caused peo-
ple to become more attuned to their own spontaneous biases and every-
day instances of discrimination and that these experiences, coupled with
increased caring, may have created ever-rising levels of concern. Future
studies should increase the frequency of measurement of both concern
and discrepancies to determine the precise time-course of the changes
on these variables. Such studies should also measure concern and dis-
crepancies immediately after the administration of the intervention to
determine whether the impact on these variables occurs immediately
after the intervention or only after people have had time to observe sub-
tle discrimination within themselves and in their environment.
Interestingly, none of the explicit variables measured at baseline
served as moderators of the effect of the intervention on implicit bias.
This is somewhat surprising given that some of the variables tested,
such as internal and external motivations to respond without prejudice
and prejudice-relevant discrepancies, have been previously implicated
in various processes that should affect receptivity to bias-reducing inter-
ventions (Monteith, 1993; Plant & Devine, 2009). We speculate that our
intervention, which was both interactive and narrated, created little op-
portunity for unmotivated or unaware participants to tune out the edu-
cation andtraining components. Hence, people who may not otherwise
have been engaged by the intervention may have, despite the lack of a
priori personal or external motivation to respond without prejudice,
found themselves compelled by the content of the intervention and
therefore began to make efforts to regulate their bias.
In contrast to the baseline explicit measures, one explicit measure at
collected at week 2, concern about discrimination, didemerge as a mod-
erator of the intervention's effect. This nding crucially implicates con-
cern about discrimination in the bias-reducing process. Given that week
2 concern, and not baseline concern, moderated the effect of the inter-
vention, this nding also suggests that concern is not important when
bias-reducing strategies are learned, but that it is important afterwards,
when people become aware of personal or societal expressions of bias
and must translate their knowledge of bias-reducing strategies into ac-
tion. Given the importance of concern in predicting who reduced their
bias, this nding highlights the need to explore what aspects of the mul-
tifaceted intervention were responsible for increasing concern. Educa-
tion might be essential for evoking concern, but other components of
the intervention may be necessary as well (e.g., being situationally
aware of one's implicit bias prior to the narrated slide show). Exploring
these issues will enable the design of more effective interventionsin the
future and help us understand the precise psychological process that
implicates concern in bias reduction.
Our ndings regarding the free-response descriptions of strategy
use and reported likelihood of use suggest a potential process re-
sponsible for the initial reduction in implicit bias and the mainte-
nance of that reduction use and anticipated use of the strategies.
The use of implementation-related word stems in describing strate-
gy use at week 2, but not awareness or motivationally related word
stems, predicted reduced implicit bias at week 4. This suggests that
the use and practice of strategies in the period immediately follow-
ing the intervention is particularly important to initial bias reduc-
tion. The fact that word stem usage at week 2 did not predict
implicit bias at week 8 suggests that the factors involved in initial
strategy deployment are different from the factors involved in the
maintenance of decreased implicit bias. Likelihood predicted reduc-
tion in implicit bias at both week 4 and week 8, even after controlling
for all the other strategy measures. This suggests that the interven-
tion generates intentions to use the strategies that are crucial to
the maintenance of the bias-reducing process. Future work should
continue to explore how and under what circumstances the strate-
gies are integrated into people's lives to effect change in implicit bias.
The word frequency ndings, which implicated overall strategy
use as being important for reducing implicit bias, do not reveal
whether people generally prefer one strategy over another or wheth-
er one particular strategy is more effective than another at producing
the various outcomes of the intervention. Because the different strat-
egies exert their effects through different psychological mechanisms,
and because the strategies are likely used in different situations, they
might have specialized effects on outcomes relevant to the regulation
of implicit bias. For example, the stereotype replacement technique
requires becoming situationally aware of the fact that one has or is
likely to have a biased response. Consequently, frequent use of the
stereotype replacement technique might lead to increased discrepan-
cies as people become chronically sensitive into the fact that they re-
spond with stereotypic biases. In contrast, the perspective taking
technique requires experiencing the world from the perspective of a
stigmatized person. As people use this technique, they might come
to better understand the consequences of subtle discrimination for
outgroups, thereby becoming more concerned about discrimination.
Unfortunately, the present study did not include precise quantitative
indicators of use of the ve strategies taught to the participants.
Collecting more precise indicators of strategy use may help to delin-
eate the extent to which use of specic strategies relates to specic
outcomes and may shed light on the mechanisms underlying the
regulation of implicit bias.
Thewordfrequencyandlikelihoodndings, combined with the nd-
ings about change in implicit bias, personal awareness, and concern, sup-
port the prejudice habit model and other dual-process models that
1276 P.G. Devine et al. / Journal of Experimental Social Psychology 48 (2012) 12671278
identify effort as necessary for implicit bias reduction (Devine, 1989;
Devine & Monteith, 1993; Smith & DeCoster, 2000; Strack & Deutsch,
2004). The support for the prejudice habit model could be strengthened
through obtaining measures of the specic behavioral process (e.g., use
of a particular strategy or set of strategies) required to produce change
in implicit bias, concern, and personal awareness. Nevertheless, we be-
lieve that our ndings provide an important demonstration that an inter-
vention can engage the long-term regulation of implicit bias, as well as
some preliminary evidence that the regulation occurs through the use
of strategies.
Future studies will need to establish the specic behavioral, cogni-
tive, affective, and neural mechanisms through which this intervention
exerts its effects. In addition to measuring when, where, and with what
frequencies people use the various strategies, future studies could, for
example, use multinomial modeling (Conrey, Sherman, Gawronski,
Hugenberg, & Groom, 2005; Payne, 2001) and behavioral neuroscience
(Cunningham et al., 2004; Phelps et al., 2000) to determine whether the
intervention changed negative associations about Black people or led to
more efcient control over automatic bias. The identication of specic
mechanisms will lead to a better understanding of how, why, and under
what circumstances the intervention will be effective.
Our intervention was multifaceted by design. This decision was
guided by the fact that (1) we wished to test whether it was possible
to engage a complex self-regulatory process involved in voluntary ef-
forts to decrease implicit bias over time and (2) it was not possible to
specify a priori which elements of the intervention (e.g., feedback re-
garding one's personal level of implicit bias, education about the nature
and consequences of implicit bias, training regarding strategies to re-
duce bias and opportunities to report on strategy use, questionnaires
that subtly remind the participants about the material presented in
the intervention) would be necessary or sufcient to engage the regula-
tory process(Howell, 1982; Kaufman, 2003; Prochaska & Velicer, 1997).
The various components of the intervention were intended to increase
awareness of bias, increase concern about discrimination, and teach
strategies that reduce bias as well as assess strategy use.
Though effective overall, the complexity of the intervention results in
ambiguity regarding which components are responsible for its various ef-
fects. For example, as shown in our control group, feedback about the
presence of implicit bias is not sufcient to trigger bias reduction, but it
may necessary. Education may play a specialized role in increasing aware-
ness and concern, but both education and training may be necessary to
produce changes in implicit bias. Additionally, our effort to assess strategy
use between the laboratory sessions may have been crucial for the
intervention's effectiveness by stimulating participants to think about
the strategies and how they could be applied in their everyday
experiences.
Although the complexity of the intervention brings ambiguity in the
interpretation of theeffects of the intervention, it is alsolikely that there
is no single magic bulletthat, by itself, prompts the regulation of im-
plicit bias and the multifarious changes in concern and awareness
such self-regulation brings. Instead, several components likely work in
combination to prompt situational awareness of one'sbias and translate
that awareness into chronic awareness, concern, and self-regulatory ef-
fort. Future studies that dismantle the intervention by systematically
manipulating the intervention's components will help identify which
components of the intervention are necessary and sufcient to produce
its distinct effects.
The intervention had a lasting effect on one measure of implicit race
bias. Although many scholars have argued that implicit bias plays a pivot-
al role in the perpetuation of discrimination (Bargh, 1999; Devine, 1989;
Fiske, 1998; Smedley et al., 2003), it will be important to demonstrate that
reductions in implicit bias lead to reductions in discriminatory outcomes
(e.g., interracial interaction quality, interview and hiring decisions, treat-
ment in health settings).
In sum, this study presents the rst intervention of its kind, one that,
using a randomized controlled design, produces a reduction in implicit
race bias that endures for at least two months. Our data provide evi-
dence demonstrating the power of the conscious mind to intentionally
deploy strategies to overcome implicit bias. As such, these ndings raise
the hope of solving a problem that has long vexed social scientists
how to reduce race-based discrimination. By empowering people to
break the prejudice habit, this study takes an important step toward re-
solving the paradox of ongoing discrimination in a nation founded on
the principle of equality.
Appendix A. Concern measure
Please indicate the degree to which you agree or disagree with each of
the following statements using the scale below.
1. I'm not personally concerned about discrimination against Blacks.
2. People need to stop focusing so much time and energy worrying
about racial discrimination.
3. People make more fuss about discrimination against Blacks than is
necessary.
4. I consider racial discrimination to be a serious social problem.
References
Amodio, D. M., Devine, P. G., & Harmon-Jones, E. (2007). A dynamic model ofguilt: Impli-
cations for motivation and self-regulation in the context of prejudice. Psychological
Science,18,524530, http://dx.doi.org/10.1111/j.1467-9280.2007.01933.x.
Bargh, J. (1999). The cognitive monster: The case against the controllability of auto-
matic stereotype effects. In S. Chaiken, & Y. Trope (Eds.), Dual-process theories in so-
cial psychology (pp. 361382). New York: Guilford Press.
Beck, A. T., & Alford, B. A. (2009). Depression: Causes and treatment (2nd ed.). Baltimore,
MD: University of Pennsylvania Press.
Bertrand, M., & Mullainathan, S. (2004). Are Emily and Greg more employable than
Lakisha and Jamal? A eld experiment on labor market discrimination. American
Economic Review,94, 9911013, http://dx.doi.org/10.1257/0002828042002561.
Blair, I. V. (2002). The malleability of automatic stereotypes and prejudice. Personality and So-
cial Psychology Review,6, 242261, http://dx.doi.org/10.1207/S15327957PSPR0603_8.
Blair, I. V., Ma, J. E., & Lenton, A. P. (2001). Imagining stereotypes away: The moderation
of implicit stereotypes through mental imagery. Journal of Personality and Social
Psychology,81, 828841, http://dx.doi.org/10.1037/0022-3514.81.5.828.
Blanton, H., Jaccard, J., Klick, J., Mellers, B., Mitchell, G., & Tetlock, P. E. (2009). Strong
claims and weak evidence: Reassessing the predictive validity of the IAT. Journal
of Applied Psychology,94, 567582, http://dx.doi.org/10.1037/a0014665.
Bradford, L. D., Newkirk, C., & Holden, K. B. (2009). Stigma and mental health in African
Americans. In R. L. Braithwaite, S. E. Taylor, & H. M. Treadwell (Eds.), Health issues
in the black community. San Francisco, CA: Jossey-Bass.
Brewer, M. (1988). A dual-process model of impression formation. In T. Srull, & R. Wyer
(Eds.), Advances in social cognition (pp. 136). Hillsdale N.J.: Erlbaum Associates.
Brigham,J. C. (1993). College students' racial attitudes. Journal of Applied Social Psychology,
23,19331967, http://dx.doi.org/10.1111/j.1559-1816.1993.tb01074.x.
Conrey, F. R., Sherman, J. W., Gawronski, B., Hugenberg, K., & Groom, C. J. (2005). Sep-
arating multiple processes in implicit social cognition: The quad model of implicit
task performance. Journal of Personality and Social Psychology,89, 469487, http:
//dx.doi.org/10.1037/0022-3514.89.4.469.
Cox, W. T. L., Abramson, L. Y., Devine, P. G., & Hollon, S. D. (2001). Stereotypes, Preju-
dice, and Depression: The Integrated Perspective. Perspectives on Psychological Sci-
ence,7,http://dx.doi.org/10.1177/1745691612455204.
Cunningham, W. A., Preacher, K. J., & Banaji, M. R. (2001). Implicit attitude measures:
Consistency, stability, and convergent validity. Psychological Science,12, 163170,
http://dx.doi.org/10.1111/1467-9280.00328.
Cunningham, W. A., Raye, C. L., & Johnson, M. K. (2004). Implicit and explicit eval-
uation: fMRI correlates of valence, emotional intensity, and control in the pro-
cessing of attitudes. Journal of Cognitive Neuroscience,16, 17171729, http:
//dx.doi.org/10.1162/0898929042947919.
Dasgupta, N., & Asgari, S. (2004). Seeing is believing: Exposure to counterstereotypic women
leaders and its effect on the malleability of automatic gender stereotyping. Journal of Ex-
perimental Social Psychology,40, 642658, http://dx.doi.org/10.1016/j.jesp. 2004.02.003.
Dasgupta, N., & Greenwald, A. G. (2001). On the malleability of automatic atti-
tudes: Combating automatic prejudice with images of admired and dis liked in-
dividuals. Journal of Personality and Social Psychology,81, 800814, http:
//dx.doi.org/10.1037/0022-3514.81.5.800.
Devine, P. G. (1989). Stereotypes and prejudice: Their automatic and controlled
components. Journal of Personality and Social Psychology,56,518, http:
//dx.doi.org/10.1037/0022-3514.56.1.5.
1 2 3456789 10
Strongly strongly
Disagree agree
1277P.G. Devine et al. / Journal of Experimental Social Psychology 48 (2012) 12671278
Devine, P. G., & Monteith, M. J. (1993). The role of discrepancy-associated affect in prej-
udice reduction. Affect, cognition, and stereotyping: Interactive processes in group
perception (pp. 317344). San Diego, CA: Academic Press.
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, 817830,
http://dx.doi.org/10.1037/0022-3514.60.6.817.
Epstein, S. (1994). Integration of the cognitive and the psychodynamic unconscious.American
Psychologist,49,709724, http://dx.doi.org/10.1037/0003-066X.49.8.709.
Feinerer, I., Hornik, K., & Meyer, D. (2008). Text mining infrastructure in R. Journal of Statistical
Software,25,154.
Fellbaum, C. (1998). WordNet: An electronic lexical database. Cambridge, MA: MIT
Press.
Fiske, S. (1998). Stereotyping, prejudice, and discrimination. In S. Fiske, D. Gilbert, & L.
Gardner (Eds.), The handbook of social psychology (pp. 357411). (4th ed.). Boston
[etc.]: The McGraw-Hill.
Fiske, S. T., & Neuberg, S. L. (1990). A continuum of impression formation, from
category-based to individuating processes: Inuences of information and motiva-
tion on attention and interpretation. In M. P. Zanna (Ed.), Advances in Experimental
Social Psychology,Vol. 23. (pp. 174)San Diego, CA: Academic Press.
Gaertner, S., & Dovidio, J. F. (1986). The aversive form of racism. In J. F. Dovidio, & S. Gaertner
(Eds.), Prejudice, discrimination, and racism (pp. 6189). Orlando: Academic Press.
Galinsky, A. D., & Moskowitz, G. B. (2000). Perspective-taking: Decreasing stereotype
expression, stereotype accessibility, and in-group favoritism. Journal of Personality
and Social Psychology,78, 708724, http://dx.doi.org/10.1037/0022-3514.78.4.708.
Green, A. R., Carney, D. R., Pallin, D. J., Ngo, L. H., Raymond, K. L., Iezzoni, L. I., et al.
(2007). Implicit bias among physicians and its prediction of thrombolysis decisions
for Black and White patients. Journal of General Internal Medicine,22, 12311238,
http://dx.doi.org/10.1007/s11606-007-0258-5.
Greenwald, A. G., McGhee, D. E., & Schwartz, J. L. (1998). Measuring individual differ-
ences in implicit cognition: The Implicit Association Test. Journal of Personality
and Social Psychology,74, 14641480.
Greenwald, A. G., Nosek, B. A., & Banaji, M. R. (2003). Understanding and using the Im-
plicit Association Test: I. An improved scoring algorithm. Journal of Personality and
Social Psychology,85, 197216, http://dx.doi.org/10.1037/0022-3514.85.2.197.
Greenwald, A. G., Poehlman, T. A., Uhlmann, E. L., & Banaji, M. R. (2009). Un-
derstanding and using the Implicit Association Test: III. Meta-analysis of
predictive validity. Journal of Personality and Social Psychology,97, 1741,
http://dx.doi.org/10.1037/a0015575.
Hofmann, W., Gawronski, B., Gschwendner, T., Le, H., & Schmitt, M. (2005). A meta-analysis
on the correlation between the implicit association test and explicit
self-report measures. Personality & Social Psychology Bulletin,31, 13691385,
http://dx.doi.org/10.1177/0146167205275613.
Howell, W. S. (1982). The empathic communicator. Prospect Heights, IL: Waveland.
Kaufman, D. M. (2003). ABC of learning and teaching in medicine: Applying educational
theory in practice. BMJ,326,213216, http://dx.doi.org/10.1136/bmj.326.7382.213.
Kim, D. (2003). Voluntary controllability of the Implicit Association Test (IAT). Social
Psychology Quarterly,66, 83, http://dx.doi.org/10.2307/3090143.
McConnell, A. R., & Leibold, J. M. (2001). Relations among the implicit association test,
discriminatory behavior, and explicit measures of racial attitudes. Journal of
Experimental Social Psychology,37,435442, http://dx.doi.org/10.1006/jesp.
2000.1470.
Mitchell, T. L., Haw, R. M., Pfeifer, J. F., & Meissner, C. A. (2005). Racial bias in mock juror
decision-making: A meta-analytic review of defendant treatment. Law and Human
Behavior,29, 621637, http://dx.doi.org/10.1007/s10979-005-8122-9.
Monteith, M. J. (1993). Self-regulation of prejudiced responses: Implications for prog-
ress in prejudice-reduction efforts. Journal of Personality and Social Psychology,65,
469485, http://dx.doi.org/10.1037/0022-3514.65.3.469.
Monteith, M. J., & Voils, C. I. (1998). Proneness to prejudiced responses: Toward under-
standing the authenticity of self-reported discrepancies. Journal of Personality and
Social Psychology,75, 901916, http://dx.doi.org/10.1037/0022-3514.75.4.901.
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,
395417, http://dx.doi.org/10.1521/soco.19.4.395.20759.
Nosek, B. A., Banaji, M., & Greenwald, A. G. (2002). Harvesting implicit group attitudes and be-
liefs from a demonstration web site. Group Dynamics: Theory, Research, and Practice,6,
101115, http://dx.doi.org/10.1037//1089-2699.6.1.101.
Paluck, E. L., & Green, D. P. (2009). Prejudice reduction: What works? A review and as-
sessment of research and practice. Annual Review of Psychology,60, 339367, http:
//dx.doi.org/10.1146/annurev.psych.60.110707.163607.
Payne, B. K. (2001). Prejudice and perception: The role of automatic and controlled
processes in misperceiving a weapon. Journal of Personality and Social Psychology,
81(2), 181192, http://dx.doi.org/10.1037/0022-3514.81.2.181.
Pettigrew, T. F. (1998). Integroup contact theory. Annual Review of Psychology,49,
6585, http://dx.doi.org/10.1146/annurev.psych.49.1.65.
Pettigrew, T. F., & Tropp, L. R. (2006). A meta-analytic test of intergroup contact
theory. Journal of Personality and Social Psychology,90, 751783, http:
//dx.doi.org/10.1037/0022-3514.90.5.751.
Phelps,E.A.,O'Connor,K.J.,Cunningham,W.A.,Funayama,E.S.,Gatenby,J.C.,
Gore,J.C.,etal.(2000).Performanceonindirectmeasuresofraceevaluation
predicts amygdala activation. Journal of Cognitive Neuroscience,12, 729738,
http://dx.doi.org/10.1162/089892900562552.
Plant, E. A., & Devine, P. G. (1998). Internal and external motivation to respond
without prejudice. Journal of Personality and Social Psychology,75, 811832,
http://dx.doi.org/10.1037/0022-3514.75.3.811.
Plant, E. A., & Devine, P. G. (2009). The active control of prejudice: Unpacking the inten-
tions guiding control efforts. Journal of Personality and Social Psychology,96,
640652, http://dx.doi.org/10.1037/a0012960.
Prochaska, J. O., & Velicer, W. F. (1997). The transtheoretical model of health be-
havior change. American Journal of Health Promotion,12,3848, http:
//dx.doi.org/10.4278/0890-1171-12.1.38.
Rubin, D. (1987). Multiple imputation for nonresponse in surveys. New York: Wiley.
Rudman, L. A., Ashmore, R. D., & Gary, M. L. (2001). Unlearningautomatic biases: The
malleability of implicit prejudice and stereotypes. Journal of Personality and Social
Psychology,81, 856868, http://dx.doi.org/10.1037/0022-3514.81.5.856.
Schuman, H., Steeh, C., Bobo, L., & Krysan, M. (1997). Racial attitudes in America: Trends
and interpretations (Rev. ed.). Cambridge, MA: Harvard University Press.
Smedley, B., Stith, A., & Nelson, A. (Eds.). (2003). Unequal treatment: Confronting racial
and ethnic disparities in health care. Washington D.C.: National Academy Press.
Smith, E. R., & DeCoster, J. (2000). Dual-process models in social and cognitive psychology:
Conceptual integration and links to underlying memory systems. Personality and Social
Psychology Review,4, 108131, http://dx.doi.org/10.1207/S15327957PSPR0402_01.
Steele, C. M. (1997). A threat in the air: How stereotypes shape intellectual identity
and performance. American Psychologist,52, 613629, http://dx.doi.org/
10.1037/0003-066X.52.6.613.
Strack, F., & Deutsch, R. (2004). Reective and impulsive determinants of so-
cial behavior. Personality and Social Psychology Review,8, 220247, http:
//dx.doi.org/10.1207/s15327957pspr0803_1.
Van Breukelen, G. J. P. (2006). ANCOVA versus change from baseline had more
power in randomized studies and more bias in nonrandomized studies.
Journal of Clinical Epidemiology,59, 920925, http://dx.doi.org/10.1016/
j.jclinepi.2006.02.007.
Vontress, C., Woodland, C., & Epp, L. (2007). Cultural dysthymia: An unrecognized dis-
order among African Americans? Journal of Multicultural Counseling and Develop-
ment,35, 130141.
1278 P.G. Devine et al. / Journal of Experimental Social Psychology 48 (2012) 12671278
... In fact, women led, and co-authored articles were favoured by referees and editors [18]. Nonetheless, some studies have demonstrated that implicit bias training may lead to modifying behavior [19][20][21][22]. Thus, EDI training and other resources, such as unconscious bias [23,24] and indigenous cultural competency training [25,26], should be easily accessible and completed by the editorial teams and authors alike. ...
Article
Full-text available
There are growing bodies of evidence demonstrating the benefits of equity, diversity, and inclusion (EDI) on academic and organizational excellence. In turn, some editors have stated their desire to improve the EDI of their journals and of the wider scientific community. The Royal Society of Chemistry established a minimum set of requirements aimed at improving EDI in scholarly publishing. Additionally, several resources were reported to have the potential to improve EDI, but their effectiveness and feasibility are yet to be determined. In this commentary we suggest six approaches, based on the Royal Society of Chemistry set of requirements, that journals could implement to improve EDI. They are: (1) adopt a journal EDI statement with clear, actionable steps to achieve it; (2) promote the use of inclusive and bias-free language; (3) appoint a journal’s EDI director or lead; (4) establish a EDI mentoring approach; (5) monitor adherence to EDI principles; and (6) publish reports on EDI actions and achievements. We also provide examples of journals that have implemented some of these strategies, and discuss the roles of peer reviewers, authors, researchers, academic institutes, and funders in improving EDI.
... First, the agency might consider their organizational diversity and ways to foster equity, diversity and inclusion at an organizational level (see Onyeador et al., 2021). Additionally, it is highly encouraged that staff participate in ongoing trainings to help address race/ethnicity related implicit biases or stereotypes (Devine et al., 2012;FitzGerald et al., 2019;Fix, 2020). In line with these recommendations, the integration of culturally/racially responsive and competent policies, programing, and services to help meet the needs and responsivity factors of youth is paramount (see Ancis, 2004;National Council of La Raza, 2007;National Juvenile Justice Network, 2014;Willison et al., 2010;Youth Collaboratory, 2019). ...
Thesis
Justice-involved youth are exposed to adverse childhood experiences (ACEs) at higher rates than youth in the general public, highlighting the importance of addressing childhood trauma and adversity in juvenile justice settings. A majority of ACEs research has focused on the general population and has demonstrated the long lasting negative impact of ACEs, on mental health, physical health, and engagement in health risk behaviors. Both gender and racial/ethnic differences have been identified in ACEs literature, suggesting that not all groups in society have the same likelihood of experiencing ACEs. Additionally, ACEs may also impact individuals from racial/ethnic or gender groups differently, resulting in variable outcomes. In comparison to the ACEs literature among the general public, little research has examined ACEs among justice involved youth, and even fewer studies have examined gender and racial/ethnic differences in these settings. A historical account of gender and racial/ethnic discrimination within the juvenile justice system, coupled with the feminist pathways perspective within an intersectional context, illustrates gendered racial/ethnic differences regarding pathways into the system and ongoing discrimination. To advance the ACEs literature, this dissertation explores the prevalence of ACEs as well as the relationship between ACEs, behavioral factors associated with delinquency, and recidivism within gendered racial/ethnic groups of justice-involved youth. The findings of the current study demonstrate the importance of accounting for both gender and race/ethnicity, as few studies have done so. Overall, the findings were mixed in relation to the prior literature and highlight the need for more research in this area, as few conclusions can be drawn from the current study’s findings. While more research is needed, broad policy implications are drawn from this study to help guide equitable assessment and treatment/services of trauma among justice-involved youth.
... Theory on prejudice reduction has suggested that becoming aware of biases is not enough to reduce them, and one must also be aware of and motivated to avoid the social consequences of their biases (Plant & Devine, 2009). Including mindful attention alongside interventions that bring awareness to the consequences of bias (e.g., stereotype replacement and counter-stereotypic imaging (Devine et al., 2012) could show potential in reducing defensive responses to learning that one is biased. ...
Article
Full-text available
Objectives Mindful attention deployment has been found to have practical benefits for a range of interpersonal outcomes including prosocial action and emotion. Recently, theory has posited that contemplative training that incorporates mindful attention may enhance intergroup compassion. Methods Here, we conduct a selective narrative review, drawing on the Buddhist concept skillful means to ask if mindful attention deployment presents an optimal starting point for intergroup compassion and action. Results An interdisciplinary theoretical framework is presented, which suggests that mindful attention dismantles common intrapsychic challenges to intergroup prosociality. Empirical research is described concerning cause and effect relationships between mindfulness and several outgrowths of intergroup prosociality. Specifically, mindfulness promotes basic social cognitive processes that allow intergroup prosociality to flourish. Conclusions While this research is promising, to date, the science on this topic has been limited to individual-level outgrowths of mindfulness practice. Discussion focuses on the future of mindfulness research in intergroup prosociality and calls for an integrative approach situating mindful attention deployment within social (and other) psychological interventions to enhance intergroup compassion.
... Solving social issues requires durable attitude changes, yet nearly all research on implicit attitude interventions has been short-term (Forscher et al., 2019). Despite temporary malleability, there is little evidence for interventions that create changes lasting more than a few days (Devine et al., 2012; but see prejudice habit-breaking intervention, Forscher et al., 2017). This underscores a need to investigate the possibility of enduring, societal-scale changes in attitudes. ...
Article
Recently, interest in aggregate and population-level implicit and explicit attitudes has opened inquiry into how attitudes relate to sociopolitical phenomenon. This creates an opportunity to examine social movements as dynamic forces with the potential to generate widespread, lasting attitude change. Although collective action remains underexplored as a means of reducing bias, we advance historical and theoretical justifications for doing so. We review recent studies of aggregate attitudes through the lens of social movement theory, proposing movements as a parsimonious explanation for observed patterns. We outline a model for conceptualizing causal pathways between social movements and implicit and explicit attitudes among participants, supporters, bystanders, and opponents. We identify six categories of mechanisms through which movements may transform attitudes: changing society; media representations; intergroup contact and affiliation; empathy, perspective-taking, and reduced intergroup anxiety; social recategorization; and social identification and self-efficacy processes. Generative questions, testable hypotheses, and promising methods for future work are discussed.
Article
This paper examines an intervention using experiential learning and simulations in a virtual world that can promote social proximity, tolerance, and cooperation in diverse societies. The participants in the study were 125 Jewish and Arab students living in Israel. A mixed linear model for repeated measures analysis that included time of measurement (pre and post), ethnicity, and students' age as independent variables revealed a main effect for time for most social groups included in this study; that is, the participants reported more social proximity to other groups after the course, including groups not studied in the course (the LGBTQ community and people of color). The results of the study suggest that experiential learning has considerable potential in the field of education to help students question their prejudices, experience being someone else, and ultimately feel social proximity for the other, thus reducing stigmas and racism.
Article
Postpartum pain is common, yet patient experiences and clinical management varies greatly. In the United States, pain-related expectations and principles of adequate pain management have been framed within established norms of Western clinical medicine and a biomedical understanding of disease processes. Unfortunately, this positioning of postpartum pain and the corresponding coping strategies and pain treatments is situated within cultural biases and systemic racism. This paper summarizes the history and existing literature that examines racial inequities in pain management to propose guiding themes and suggestions for innovation. This work is critical for advancing ethical practice and establishing more effective care for all patients.
Article
Importance: Linguistic differences suggestive of gender bias have been detected in letters of recommendation (LOR) for female and male residency and fellowship program applicants within multiple medical specialties. Objective: The aim of the study was to determine whether linguistic differences exist in LOR for female and male physicians applying to female pelvic medicine and reconstructive surgery (FPMRS) fellowship. Study design: A retrospective analysis of FPMRS fellowship applications submitted to a university-affiliated academic center from 2010 to 2020 was performed. Linguistic Inquiry and Word Count, a text analysis software, was used to characterize the linguistic content of letters. Multivariable analysis was used to compare letter characteristics with applicant and letter writer demographics. Results: Of 306 applications reviewed, 221 (72.2%) applicants were female and 85 (27.8%) were male. Of the 1,062 letters analyzed, 457 (43.0%) were written by female letter writers, 586 (55.2%) by males, and 19 (1.8%) were a combination. Multivariable analysis controlling for race, Step 1 score, and letter writer gender demonstrated more frequent use of affiliation words for female applicants compared with males (3.1% ± 0.3 vs. 2.9% ± 0.3, P = 0.02). No additional differences were noted in average letter length or all other linguistic categories analyzed. Multiple differences were detected between writing styles of female and male letter writers, including average letter length, use of multiple word categories, and use of communal (relationship-oriented) language. Data were stratified into 2-year periods and no longitudinal trends in linguistic differences were detected. Conclusions: No linguistic differences, suggestive of gender bias, were found between female and male applicants to FPMRS fellowship.
Article
Refugees can experience elevated levels of psychological distress upon resettlement, although disparate outcomes over time are expected. The current study modeled trajectories of changes in distress over a 5-year period among resettled refugees and sought to explicate post-settlement factors that influence distress over time. A large-scale sample of refugees resettled in Australia (N = 2,399) were tracked over a 5-year period, completing measures of psychological distress at each wave and initial risk and protective factors immediately after resettlement. A latent class growth analysis conducted on distress found four unique classes characterised by (1) resilient levels of distress, (2) consistent clinical distress, (3) recovering levels of distress, and (4) deteriorating distress. Lower perceived discrimination and greater positive context of reception predicted membership to the resilient group and differentiated the recovering and deteriorating groups. Further, lower ingroup social support predicted membership to the clinically distressed group relative to all others. We conclude by echoing calls to strengthen community support for refugees and promote ingroup ties, particularly among those who are the most vulnerable.
Article
Research on implicit stereotypes has raised important questions about an individual's ability to moderate and control stereotypic responses. With few strategies shown to be effective in moderating implicit effects, the present research investigates a new strategy based on focused mental imagery. Across 5 experiments, participants who engaged in counterstereotypic mental imagery produced substantially weaker implicit stereotypes compared with participants who engaged in neutral, stereotypic, or no mental imagery. This reduction was demonstrated with a variety of measures, eliminating explanations based on response suppression or shifts in response criterion. Instead, the results suggest that implicit stereotypes are malleable, and that controlled processes, such as mental imagery, may influence the stereotyping process at its early as well as later stages.
Article
The present research suggests that automatic and controlled intergroup biases can be modified through diversity education. In 2 experiments, students enrolled in a prejudice and conflict seminar showed significantly reduced implicit and explicit anti-Black biases, compared with control students. The authors explored correlates of prejudice and stereotype reduction. In each experiment, seminar students' implicit and explicit change scores positively covaried with factors suggestive of affective and cognitive processes, respectively. The findings show the malleability of implicit prejudice and stereotypes and suggest that these may effectively be changed through affective processes.
Article
Respondents at an Internet site completed over 600,000 tasks between October 1998 and April 2000 measuring attitudes toward and stereotypes of social groups. Their responses demonstrated, on average, implicit preference for White over Black and young over old and stereotypic: associations linking male terms with science and career and female terms with liberal arts and family. The main purpose was to provide a demonstration site at which respondents could experience their implicit attitudes and stereotypes toward social groups. Nevertheless, the data collected are rich in information regarding the operation of attitudes and stereotypes, most notably the strength of implicit attitudes, the association and dissociation between implicit and explicit attitudes, and the effects of group membership on attitudes and stereotypes. (PsycINFO Database Record (c) 2013 APA, all rights reserved)
Article
Three studies tested basic assumptions derived from a theoretical model based on the dissociation of automatic and controlled processes involved in prejudice. Study 1 supported the model's assumption that high- and low-prejudice persons are equally knowledgeable of the cultural stereotype. The model suggests that the stereotype is automatically activated in the presence of a member (or some symbolic equivalent) of the stereotyped group and that low-prejudice responses require controlled inhibition of the automatically activated stereotype. Study 2, which examined the effects of automatic stereotype activation on the evaluation of ambiguous stereotype-relevant behaviors performed by a race-unspecified person, suggested that when subjects' ability to consciously monitor stereotype activation is precluded, both high- and low-prejudice subjects produce stereotype-congruent evaluations of ambiguous behaviors. Study 3 examined high- and low-prejudice subjects' responses in a consciously directed thought-listing task. Consistent with the model, only low-prejudice subjects inhibited the automatically activated stereotype-congruent thoughts and replaced them with thoughts reflecting equality and negations of the stereotype. The relation between stereotypes and prejudice and implications for prejudice reduction are discussed.