Content uploaded by David S. Yeager
Author content
All content in this area was uploaded by David S. Yeager on Aug 19, 2020
Content may be subject to copyright.
Loss of Institutional Trust Among Racial and Ethnic Minority Adolescents: A
Consequence of Procedural Injustice and a Cause of Life-Span Outcomes
David S. Yeager
University of Texas at Austin
Valerie Purdie-Vaughns
Columbia University
Sophia Yang Hooper
University of Texas at Austin
Geoffrey L. Cohen
Stanford University
This research tested a social-developmental process model of trust discernment. From sixth to eighth grade,
White and African American students were surveyed twice yearly (ages 11–14; Study 1, N=277). African
American students were more aware of racial bias in school disciplinary decisions, and as this awareness
grew it predicted a loss of trust in school, leading to a large trust gap in seventh grade. Loss of trust by
spring of seventh grade predicted African Americans’subsequent discipline infractions and 4-year college
enrollment. Causality was confirmed with a trust-restoring “wise feedback”treatment delivered in spring of
seventh grade that improved African Americans’eighth-grade discipline and college outcomes. Correlational
findings were replicated with Latino and White students (ages 11–14; Study 2, N=206).
I am marked by old codes, which shielded me in
one world and then chained me in the next.—Ta-
Nehisi Coates (2015), an African American father, in
a letter to his adolescent son.
A key developmental challenge in contemporary
society is to learn how to monitor, assess, and reas-
sess the trustworthiness of institutions and ulti-
mately make judgments about whether compliance
with institutional policies is warranted (Purdie-
Vaughns, Steele, Davies, Ditlmann, & Crosby, 2008;
Spencer, 2006; Tyler & Blader, 2003). Adolescents
face this challenge with respect to many institu-
tions, such as police, government, healthcare, or
businesses. Here, we focus on one setting where
trust discernment comes to the fore and may have
long-term consequences: racial and ethnic minority
adolescents transitioning through middle school.
On the one hand, it can be costly for an adoles-
cent to trust an institution when it systematically
takes advantage of one’s group (Cohen & Steele,
2002; Gambetta, 1988). Indeed, in the U.S. schools,
African American and Latino youth are dispropor-
tionately subjected to mistreatment by authorities,
in the form of low expectations from teachers (see
Harber, 1998), extreme disparities in disciplinary
referrals and suspensions for minor misbehavior
(Okonofua & Eberhardt, 2015; Skiba, Michael,
Nardo, & Peterson, 2002), and even unwarranted
stops and arrests by police (see Tyler, Fagan, &
Support for this research was provided by grants from the
National Science Foundation/REESE Division (Award 0723909),
the Spencer Foundation (Award 200800068), the W. T. Grant
Foundation, the Nellie Mae Education Foundation, and National
Science Foundation Grant DRL-1109548 to the last author. Writ-
ing of the manuscript was supported by a fellowship from the
Center for Advanced Study in the Behavioral Sciences (CASBS)
and a William T. Grant Scholars award to the David S. Yeager.
The authors are grateful to Eric Bettinger for his assistance
obtaining National Student Clearinghouse data and to Hae Yeon
Lee for her analyses of the School and Staffing Survey. Data, syn-
tax, and results of alternative model specifications are available
online at osf.io/3hpu8.
Correspondence concerning this article should be addressed to
Geoffrey L. Cohen or David S. Yeager. 108 E. Dean Keeton Stop
A8000, Austin, TX, 78712-1043. Electronic mail may be sent to
glc@stanford.edu or dyeager@utexas.edu.
[Article corrected on March 31, 2017 after first online publication
on February 8, 2017: On page 659, “were white”was inserted into
the sentence, “. . .70% of teachers in minority-serving schools were
white (i.e. those with 50% or more non-white students. . .”.On
page 662, two variations of the same sentence were corrected to
read, “In these experiments, wise feedback clarified that the critical
feedback originated in the critic‘s positive motivations rather than
bias.”On page 673, a duplicate appearance of “send the”was
removed from the sentence, “Instead, they continually send the
message that their students are capable. . .”On page 674, the
reference to a Gallup poll was corrected to be “Jones, J. M. (2015,
June 19). In U.S., confidence in police lowest in 22 years. Omaha,
NE: Gallup. Retrieved from http://www.gallup.com/poll/
183704/confidence-police-lowest-years.aspx”. Additionally, gram-
matical changes were made to the article.]
©2017 The Authors
Child Development ©2017 Society for Research in Child Development, Inc.
All rights reserved. 0009-3920/2017/8802-0028
DOI: 10.1111/cdev.12697
Child Development, March/April 2017, Volume 88, Number 2, Pages 658–676
Geller, 2014). Because trust of authorities is negoti-
ated in the context of social interactions (Lewin,
Lippitt, & White, 1939), a natural adaptation to
such interactions is to withhold trust.
As illustrated by the quotation above, from Ta-
Nehisi Coates (2015) to his 14-year-old son, experi-
ences of injustice can affect one in ways that can
make it more difficult to thrive in an institution. That
is, racial and ethnic minority youth, experiencing
and perceiving bias, may generalize from specific
interactions to a mental representation of the institu-
tion as an abstract entity. Youth may then demon-
strate lower compliance with institutional policies,
accelerating a self-reinforcing cycle of punishment
and loss of trust (Crocker & Major, 1989; Fagan &
Tyler, 2005; Gregory & Weinstein, 2008; Levi, Sacks,
& Tyler, 2009; Tyler, Goff, & MacCoun, 2015).
Exacerbating the potential for mistrust is the racial
divide that characterizes most interactions between
teachers and their students in the U.S. According to
our re-analyses of the School and Staffing Survey
(Coopersmith, 2009), 70% of teachers in minority-ser-
ving schools were white (i.e. those with 50% or more
non-white students; the figure is 96% in predomi-
nately white-student-serving institutions). The ten-
dency to attribute bias is greater across racial divides
than within them (Crocker & Major, 1989); thus the
demographics of most American schools may act as
an affordance of mistrust.
Yet a loss of trust can become a liability if some
authority figures wish to enact fair policies, treat
racial or ethnic minority students with respect, and
help them to thrive (Gregory & Weinstein, 2008;
Okonofua, Walton, & Eberhardt, 2016). When stu-
dents have lost trust, they may be deprived of the
benefits of engaging with an institution, such as
positive relationships, access to resources and
opportunities for advancement, and avoidance of
punishment. Thus, racial and ethnic minority youth
may be twice harmed by institutional injustices:
They both receive the lion’s share of the initial pun-
ishment, and then may be required to psychologi-
cally adapt, through a loss of trust, in a way that
prevents them from profiting from instruction and
relationships.
This research seeks to understand this predica-
ment by applying a developmental and social-psy-
chological lens (also see Okonofua, Paunesku, &
Walton, 2016; Okonofua, Walton, et al., 2016). We
use data from an 8-year longitudinal study that
tracked White and African American students from
sixth grade until entry into college across two
cohorts (Study 1). In addition, because our theory is
that trust is an issue for any group that faces
negative stereotypes and institutional mistreatment,
we also examine data from a 1-year, cohort-sequen-
tial study of White and Latino middle school stu-
dents (Study 2).
Our approach integrates insights from develop-
mental (Killen, Mulvey, & Hitti, 2013; McKown,
2013; Olson & Dweck, 2008) and social-psychological
recursive process models (Okonofua, Walton, et al.,
2016), resulting in the model shown in Figure 1. We
document the mental representations (i.e., institu-
tional trust) that sit in the middle between objective
realities (i.e., racial and ethnic minority group differ-
ences in the perceived justice of school procedures)
and developmental outcomes (i.e., group differences
in rates of noncompliance with school rules, or in
long-term educational outcomes such as college
enrollment). We highlight how these mental repre-
sentations are shaped and then create social reality,
propelling effects forward in time.
Finally, we use a small sample, double-blind ran-
domized experiment to test causality. We illustrate
how a timely and credible show of respect from
authorities to racial and ethnic minority youth can,
during a key developmental window, set in motion
an alternative process. This more beneficial process
can culminate in improved educational behaviors
years later.
Institutional Trust Among Adults
Among adults, what causes institutional trust?
Research in social psychology (Tyler, 2006; Tyler &
Blader, 2003; Tyler et al., 2015), political science
(Levi, 1997; Levi et al., 2009), and sociology (Bryk &
Schneider, 2002; Goffman, 1963) has converged on
the conclusion that people trust an institution more
when they perceive that it is procedurally just—that
is, that it uses fair processes to make consequential
decisions—and when they believe that authorities
have personal regard for individuals served by the
institution—that is, when authorities are respectful
and have one’s best interest at heart. Notably, peo-
ple do not have to have been the personal recipient
of procedural justice or personal regard to gain or
lose trust (Tyler et al., 2014). It is enough to experi-
ence it vicariously by means of observation.
What has been missing from much previous
research, however, has been a developmental per-
spective. Indeed, an adolescent developmental lens
was a notable absence from recent influential
reviews of group disparities in crime and policing
(President’s Task Force on 21st Century Policing,
2015; Tyler et al., 2015). Our primary theoretical
contribution is to offer such a lens.
Middle School Trust Gap 659
A Developmental Process Model of the Emergence and
Consequences of Institutional Trust
Adolescence has been described as a period of
socialization into a belief about the fairness and
legitimacy of institutions (Baumrind, 1997; Fagan &
Tyler, 2005). Positive and negative encounters with
authorities—such as discipline incidents or moments
of criticism—are “teachable moments,”for good or
ill (Baumrind, 1997; Fagan & Tyler, 2005; Tyler
et al., 2014). Of course, even young children are able
to discern whether to trust another person (Vander-
bilt, Liu, & Heyman, 2011) and are aware of racial
and ethnic group membership (see Brown & Bigler,
2005). Why then might adolescence be a special per-
iod for institutional trust discernment?
To trust—or distrust—an institution, one must
notice and assign generalized meaning to authori-
ties’negative treatment of members of one’s group
and attribute it to a bias in the institution as an
abstract actor (the school, or school in general) not
just to an individual actor (a specific teacher). Sev-
eral developmental antecedents make adolescents
prepared to do this (see the top left panel in Fig-
ure 1).
Changes During Adolescence
Adolescents are increasingly capable of mentally
representing unseen, abstract policies in an institu-
tion as the cause of unfair treatment, whereas chil-
dren may be more attuned to the negative attitudes
of specific authority figures acting as individuals
(see Brown & Bigler, 2005; Spencer, 2006). Next,
adolescence is a period marked by questioning
whether to obey adults’commands (Laupa, Turiel,
& Cowan, 1995) or ignoring them altogether (for an
example from neuroscience, see Lee, Siegle, Dahl,
Hooley, & Silk, 2014). Third, adolescents may
undergo pubertal changes that can heighten a sensi-
tivity to the possibility of being disrespected by
adult authorities. The pubertal surge in testosterone
and other hormones among boys and girls (see
Peper & Dahl, 2013) is thought to increase attention
to, reactivity to, and memory for experiences of
being disrespected or made to feel low status (see
Terburg & van Honk, 2013).
Developments Specific to Minority Youth
As they mature, adolescents undergo these
changes regardless of ethnic group membership.
However, by the start of middle school, negatively
stereotyped ethnic minority adolescents are also
more likely than White peers to be racially and ethni-
cally aware—that is, to have conscious appraisals
about how different racial and ethnic groups are
evaluated and treated by the larger society (Ruble
et al., 2004; Sellers, Copeland-Linder, Martin, &
Lewis, 2006; Spencer, 2006; Swanson, Cunningham,
Youngblood, & Spencer, 2009). Long before they
get to sixth grade, children and teachers stereotype
African American children as aggressive (Sagar &
Schofield, 1980), and children take note. By early
Time
(During Adolescence)
Social
Reality
Mental
Representation
of Institution
Social
Reality
Social
Reality
Social
Reality
Social
Reality
Institutional Trust
Outcomes
Behavioral
Noncompliance
Poorer
Performance
Disidentification
With School
Developmental Processes
Lower Human
Capital
Group
Disparities in
School
Discipline
and Grades
Teacher
Criticizes
Student's
Work
Social-Psychological Processes
Trust
"Adults at
This School
Are Not Fair"
Racial
Awareness
Social-Cognitive Ability for
Abstraction
Pubertal Hormone-Sensitized
Vigilance to Fairness and Respect
Social Reality Mental Representation
Developmental Antecdents
Student Does Not
Revise, Gets Low
Grade
Low Trust "Wise
Feedback"
Example 1
Example 2
Mental
Representation
of Institution
Mental
Representation
of Institution
Mental
Representation
of Institution
Mental
Representation
of Institution
Figure 1. A social-developmental recursive process model of trust discernment during adolescence.
660 Yeager, Purdie-Vaughns, Hooper, and Cohen
adolescence, African American students report
occurrences of others perceiving them as a threat
(Sellers et al., 2006), and the overwhelming majority
are aware of negative stereotypes about their
group’s intelligence or behavior by the beginning of
middle school, even as their White peers are una-
ware (Bigler, Averhart, & Liben, 2003; McKown &
Weinstein, 2003).
Negatively stereotyped racial and ethnic
minority adolescents may therefore enter middle
school prepared to attribute unfair treatment to
group membership rather than group-irrelevant
factors and do so more readily than White peers
(Crocker & Major, 1989, McKown & Weinstein,
2003; also see Cohen & Garcia, 2005). This pre-
paredness then meets social reality in the form
of disparate punishment and remediation in
school (Okonofua & Eberhardt, 2015; see Skiba
et al., 2002). Even when students themselves are
not subjected to disparate treatment, vicarious
experiences of group members offer a window
into the potential unfairness endemic to the insti-
tution (see Bigler & Liben, 2006).
Recursive Process
Like a stone rolling down a hill that triggers an
avalanche, the loss of trust—especially among
members of negatively stereotyped minority groups
—could accumulate behavioral consequences over
time (see Roberts & Caspi, 2003; Spencer, 2006), as
illustrated in the right panel of Figure 1. For mem-
bers of stereotyped groups, disparities in treatment
in a given institution can amplify signals of unfair-
ness and disrespect (Crocker & Major, 1989, Purdie-
Vaughns et al., 2008), while obscuring signals of
fair treatment (Harber, 1998), accelerating loss of
trust, and its potential negative consequences
(Okonofua, Walton, et al., 2016).
Social-psychological process models emphasize
the importance of attributions in this feedback loop
(Cohen & Steele, 2002; Okonofua, Walton, et al.,
2016; see the left panels in Figure 1). If adolescents
experience procedural injustice or disrespect, they
may come to expect it. If they expect it, they will
tend to see it. If they see it, they will tend to
expect more (cf. Lord, Ross, & Lepper, 1979). See-
ing and expecting injustice and disrespect, nega-
tively stereotyped ethnic minority adolescents may
disengage, defy authorities, underperform, and act
out. This process may culminate in greater disci-
plinary incidents and lower academic attainment,
even among ethnic minority adolescents who
began middle school motivated and capable of
succeeding (Gregory & Weinstein, 2008; Okonofua,
Walton, et al., 2016; Spencer, 2006).
Our process model offers an alternative to deficit
models of adolescent behavior problems. It does not
posit deficiencies in executive function, cultural mis-
matches between school contexts and ethnic minority
families. Nor does it posit a developmentally deter-
ministic path in which adolescents inevitably rebel
against adult authority. These may play a role but
the causal action in our model is in the interaction
between social reality and psychology. A loss of trust
and its behavioral consequences are often reasonable
adaptations to an environment of disrespect and pro-
cedural injustice (Spencer, 2006; Swanson et al.,
2009). This adaptation can have long-term conse-
quences when left unaddressed. However, as a men-
tal representation, trust can be improved through
timely intervention (see Olson & Dweck, 2008).
Slowing a Loss of Trust Through “Wise”Strategies
That Respect Youth
What might prevent a loss of trust in adoles-
cence? “Wise”strategies can accomplish this. They
convey to students that they will be neither treated
nor judged in light of a negative stereotype but will
instead be respected and treated as a valuable indi-
vidual (Cohen & Steele, 2002; Goffman, 1963). If
such a signal was offered during sensitive periods
for trust formation, it might create accumulating
positive consequences, by means of a virtuous
recursive cycle rather than a vicious one (Yeager &
Walton, 2011).
The term “wise”was originally formulated by
Goffman (1963) in his analysis of social stigma. To
be “wise”is the act of seeing stigmatized individu-
als in their full humanity, which enables an open-
ness and honesty when one interacts with them.
Ethnographic accounts of exceptional teachers
show how wise practitioners build trust with and
motivate members of stigmatized groups (e.g., Lep-
per & Woolverton, 2002; Treisman, 1992; see Gre-
gory & Weinstein, 2008). Such teachers have been
described as “warm demanders”(Vasquez, 1988)
and “compassionate disciplinarians”(Irvine, 2003),
meaning that they are distinguished by their ability
to combine high performance standards with high
personal regard for students’well-being. One stu-
dent described a “wise”teacher this way: “When
she talk to you with seriousness, she mean [sic] it,
but then she also have a smile like ‘I’m on your
side”(Gregory & Weinstein, 2008).
In large-scale survey studies, this combination of
high standards and personal care predicted higher
Middle School Trust Gap 661
achievement for negatively stereotyped and socioe-
conomically disadvantaged youth (Gregory &
Weinstein, 2004; Shouse, 1996). Furthermore, in a
daily diary study, African American students who
reported experiencing both high expectations and
feelings of personal care were the most likely to
report that they trusted their teacher (Gregory &
Weinstein, 2008, Study 2)—regardless of the tea-
cher’s racial and ethnic group.
Research has distilled the practices of wise men-
tors to investigate them experimentally. Cohen,
Steele, and Ross (1999) showed that when African
American students were required to endure tough
criticism on an essay, they benefitted from learning
that the teacher was critical not because of bias but
because of his stated desire to hold them to a higher
standard (procedural justice) and his belief that they
were capable of meeting that higher standard (per-
sonal regard). This message—called “wise feed-
back”—increased African American college
students’motivation and reduced their attributions
of bias (Cohen et al., 1999). In these experiments,
wise feedback clarified that the critical feedback orig-
inated in the critic’s positive motivations rather than
bias. By changing attributions in this manner, the
intervention turned what might have otherwise been
seen as negative feedback into positive feedback.
This wise feedback method was later tested in a
behavioral field study that affected African Ameri-
can seventh-grade students’institutional trust and
motivation (Yeager et al., 2014). Yeager et al. (2014)
showed that a single hand-written note from stu-
dents’social studies teacher, appended to an essay
they had written, and pithily but credibly conveying
the teacher’s high standards and belief in students’
capacity to reach them, increased from 17% to 72%
the proportion of African American adolescents who
revised their critiqued essays a week later (covariate-
adjusted values). Furthermore, 2.5 months later, the
note halted the semester-over-semester decline in
trust for African American seventh graders who, at
baseline, had begun to mistrust school.
In this research (Study 1), we analyze additional
longitudinal data from the same data set originally
reported by Yeager et al. (2014; Studies 1 and 2).
We look at the implications of an experience of
wise feedback on African American adolescents’
subsequent disciplinary infractions and eventual 4-
year college enrollment.
This Research
We test whether disciplinary outcomes in school,
and the sense of procedural injustice they give rise
to, predict a loss of institutional trust for negatively
stereotyped racial and ethnic minority adolescents
(Figure 1, top left panel), and whether this loss in
turn predicts greater awareness of procedural jus-
tice, starting a feedback loop (Figure 1, right panel).
We assume but do not measure developmental
antecedents that can facilitate this process, such as
racial and ethnic awareness, social-cognitive ability,
and the start of puberty (Figure 1, top left panel).
We then test whether middle school adolescents’
trust predicts their later behavioral outcomes—their
discipline incidents in eighth grade and their even-
tual enrollment at a 4-year college (bottom right of
Figure 1).
Finally, as noted, we take advantage of a ran-
domized experimental intervention embedded in
this correlational study, to test causality. Can wise
critical feedback from a teacher, which has already
been shown to sever the effect of academic mistrust
on subsequent willingness to revise an essay and
later mistrust (Yeager et al., 2014, Studies 1 and 2),
set in motion a virtuous recursive cycle? We
explored whether a wise hand-written note from a
teacher, known to convey respect and build trust
(cf. Gregory & Weinstein, 2008), might set in
motion a process that leads to greater behavioral
compliance the subsequent year as well as greater
eventual enrollment at a 4-year college (also see
Okonofua, Paunesku, et al., 2016).
Study 1
Method
Participants
Survey sample. A total of N=277 students in
two consecutive cohorts (Fall of 2004 and 2005)
were recruited from a middle-class to lower-mid-
dle-class public middle school in the northeast
region of the United States and surveyed twice
yearly from sixth to eighth grade. The overall
school population was split evenly between African
American and White students. Twenty-two percent
were eligible for free or reduced-price lunch, an
indicator of socioeconomic status. Fifty-two percent
of the students were female; 48% were African
Americans, and 52% were White. Not all students
completed all measures, so degrees of freedom var-
ied across analyses.
Experimental subsample. In the spring of seventh
grade in each cohort, a subsample of 44 students
(88 total) with “B”or “C”grades participated in a
randomized experimental intervention in their
662 Yeager, Purdie-Vaughns, Hooper, and Cohen
social studies classrooms. The focus on “B”and
“C”students was motivated by the fact that the
intervention conveyed to students that they were
able to reach a higher standard. This message, it
was expected, would lack credibility in the eyes of
very low-performing students and seem redundant
in the eyes of very high-performing students, who
were already, by and large, reaching a higher stan-
dard (Yeager et al., 2014).
All social studies teachers were white. This
matches the modal experience for students in the
U.S., as noted previously.
In each cohort, n=11 students from each racial
group (White and African American) were assigned
either to a treatment condition or placebo control
condition, yielding 22 total per racial group per
condition. These constituted the “experimental sub-
sample”(for extensive detail on this subsample, see
Yeager et al., 2014, Studies 1–2).
Procedure
Survey sample. Students were surveyed in the
spring and fall of sixth, seventh, and eighth grades
(six total surveys). At the end of the middle school
years, data on discipline incidents and school grades
were collected from the school registrar. College
enrollment data (whether students enrolled in the
year after high school and whether it was a 4-year col-
lege or not) were obtained 6 years after middle school
through the National Student Clearinghouse (NSC).
Experimental subsample. The “wise feedback”
intervention was designed to be an antidote to mis-
trust of teachers. No intervention could fully dispel
concerns about racial bias. Instead, this treatment
was aimed to interrupt the reciprocally reinforcing
relationship between perceived bias and mistrust
(Figure 1). Because the process depends on a con-
tinual feedback loop, even a minor interruption
could have benefits that persist and compound with
time. The intervention altered students’construals
of an acute experience with the potential to increase
mistrust: the receipt of teachers’critical feedback on
an essay they had written. Such feedback could
plausibly be attributed to either negative or positive
factors, such as bias or a belief in one’s ability to
reach a higher standard (Cohen & Steele, 2002). The
treatment sought to promote the latter judgment.
In spring of seventh grade, in the context of their
social studies class, students wrote an essay about a
personal hero. Their initial draft received thorough
critical feedback from their teachers, accompanied
by whatever encouragement the teacher felt
inclined to give. Before students received their
edited essay, the research team appended one of
two notes to the essays of the participating stu-
dents. Both notes had been hand written by the tea-
cher in advance. In a placebo control condition, the
note stated, “I’m giving you these comments so that
you’ll have feedback on your paper,”which fulfills
the grammatical expectation to provide an explana-
tion but conveys no information relevant to the
trustworthiness of the criticism. In the wise feedback
condition, the note stated, “I’m giving you these
comments because I have very high expectations
and I know that you can reach them.”A blank
piece of paper covered the notes and edited essays
to keep peers and teachers blind to condition.
Over the next week, students had the opportu-
nity to revise their essays, which most did. Yeager
et al. (2014) previously reported that the randomly
assigned “wise feedback”note increased African
American students’likelihood of revising their
essays and improved the quality of their revisions.
For instance, it doubled the number of teacher-sug-
gested edits that students implemented. Further-
more, 3 months posttreatment, African American
students with low levels of baseline trust showed a
less steep decline in trust in the treatment condition
than in the control condition.
Survey Measures
School trust. Trust was assessed with six ques-
tions (e.g., “I am treated fairly by teachers and
other adults at my school,”“Students in my racial
group are treated fairly by teachers and other
adults at [school name] middle school;”1=very
much disagree,6=very much agree). Items were aver-
aged at each wave (all as>.70), with higher values
indicating greater trust.
Awareness of bias in enforcement of school poli-
cies. Students reported their belief that racial bias
intruded in school disciplinary decisions—and thus
provided a measure of procedural injustice—using
two items written specifically for this study (“If a
Black or a White [school name] student is alone in
the hallway during class time, which one would a
teacher ask for a hall pass?”and “If a Black and a
White [school name] student do something wrong,
who is more likely to get in trouble for it?”Rated on
a 5-point scale; 1 =almost always the black student
instead of the white student;3=both of them about the
same;5=almost always the white student instead of the
black student). Only 2% of all participants ever gave a
response >3, which would have indicated pro-Afri-
can American bias. Therefore, the measure is best
interpreted as perceptions of anti-African American
Middle School Trust Gap 663
bias. The two items were averaged, with higher val-
ues corresponding to a greater expectation of unjust
treatment (a>.70). We then reverse coded the com-
posite score, with higher values corresponding to a
greater expectation of procedural injustice. The
modal response was “about the same,”with over
55% of students at each wave providing a rating of 3.
Data could thus be analyzed either in their original
continuous form or dichotomized. Although we used
the former method, substantive results were
unchanged with the latter.
School records. From school records we obtained
the following data for each student: their gender, the
number of disciplinary incidents in each of the three
middle school years (range: 0–34), and grade point
averages (GPAs) in core classes (math, science,
social studies, and English or reading; range: 0 [F]
to 4.33 [A+]) for each academic quarter of each of
the three school years. As is common in this kind
of administrative data, disciplinary incidents were
nonnormal due to large numbers of students with 0
incidents and few with very many (>20). To better
normalize the distribution, high outliers were
recoded to the 98th percentile and then the overall
metric was square root transformed. However, sub-
stantive results were unchanged when this was not
done (i.e., significant results in the article were sig-
nificant in the raw data as well). In addition, for the
first cohort only, the school provided data on the
reason for each disciplinary action.
College enrollment. College enrollment data were
obtained from the NSC. NSC is a nonprofit data-
base that reports on students receiving financial aid
to both private and federal loan providers, and as a
result it tracks college enrollment and degree attain-
ment for the vast majority of college students in
America. Researchers conducting program evalua-
tion in partnership with schools, as in this case, can
obtain participating students’college enrollment
data. In the Northeast region of the United States,
where nearly all participating students attended col-
lege, the coverage rate for African American stu-
dents attending public 4-year institutions was
99.6%, at public 2-year institutions it was 99.2%,
and at private nonprofit institutions it was 97%
(Dynarski, Hemelt, & Hyman, 2013).
The primary outcome of interest was on time
enrollment at a 4-year college (1 =enrolled at a 4-year
college the year after high school; 0 =did not),
which was the appropriate measure for the middle-
achieving (B and C average) students who partici-
pated in the experiment. Exploratory analyses also
focused on whether students enrolled in college at
all; roughly 70% of students in this middle-class
school had records the year after high school, indi-
cating college enrollment. Just over two thirds of
these students enrolled in a 4-year college.
Covariates. Covariates used in all structural
equation models were as follows: students’gender,
cohort, and prior achievement. The latter was an aver-
age of students’state test scores in the first month of
sixth grade and their prior grades. To ensure that
these achievement measures represented baseline
values, the correlational analyses, which began with
data from the fall of sixth grade, used fifth-grade
GPA, whereas analyses of the experimental treat-
ment, implemented in the spring of seventh grade,
used GPA from the fall of seventh grade. College
enrollment models, which as described below used
seventh- and eighth-grade measures, also controlled
for prior (sixth grade) trust.
Results
The software Mplus was used for all structural
models. Full information maximum likelihood was
used to estimate parameters without discarding or
imputing missing data. For models predicting
school discipline (a count variable), Poisson regres-
sion was used. Additional detail on all statistical
models can be found in the Supporting Informa-
tion, and syntax is posted online (osf.io/3hpu8).
The Emergence of the Middle School Racial Trust Gap
In a growth curve model, including all students
from both racial groups, trust decreased from
Grades 6 to 8, slope: unstandardized b=0.13,
p<.001, 95% CI [0.17, 0.09]. The model fit the
data acceptably: v
2
(29) =83.45, p<.001; root mean
square error of approximation (RMSEA) =.08, 90%
CI [0.06, 0.10]; and comparative fit index (CFI) =.93.
However, trust declined faster for African Ameri-
can students, producing a racial trust gap. Figure 2
shows that there was no significant trust gap
between African American and White students in
sixth grade, but the gap emerged by fall of seventh
grade. Student race significantly predicted the slope
from sixth to eighth grades in a latent growth curve
model, unstandardized b=0.06, p=.028, 95% CI
[0.11, 0.01]. This steeper loss of trust for African
American students led to significant race-based dif-
ferences in trust by seventh and eighth grade.
Predictors of the Trust Gap
Does age alone account for the trust gap?. If age
alone—and not school experiences—accounted for
664 Yeager, Purdie-Vaughns, Hooper, and Cohen
the emergence of the trust gap across the middle
school years, then one might expect the decline in
trust over middle school to be a monotonically
decreasing linear function. However, when inspect-
ing Figure 2, the change in trust in the summer is
mostly flat or positive, whereas the change in trust
during the school year mostly slopes downward. This
is consistent with the explanation that loss of trust
emerges from experiences in school not processes tied
to chronological age or pubertal maturation alone.
To test this formally, in another growth model
conducted with all students, we estimated a “school
year”slope and a “summer”slope by fixing weights
on two latent slope variables (“School year:”fall
sixth grade =0, spring sixth grade and fall seventh
grade =1, spring seventh grade and fall eighth
grade =2, spring eighth grade =3, “Summer:”fall
sixth grade and spring sixth grade =0, fall seventh
grade and spring seventh grade =1, and fall eighth
grade and spring eighth grade =2). The model fit
Figure 2. The racial and ethnic trust gap emerges in seventh grade and is largest in spring of seventh grade in Studies 1 and 2.
Note. Values are covariate adjusted, controlling for premiddle school achievement, gender, and cohort. p-values are from t-tests;
d=Cohen’sd. Results are from independent regression models.
Middle School Trust Gap 665
the data acceptably: v
2
(22) =48.95, p<.001;
RMSEA =.07, 90% CI [0.04, 0.09]; and CFI =.97.
For all students, “school year”trust declined sig-
nificantly over time, unstandardized b=0.32,
p<.001, 95% CI [0.42, 0.22], and “summer”
trust did not, unstandardized b=0.10, p=.092,
95% CI [0.02, 0.21]. A Wald test of parameter con-
straints showed that the “school year”decline was
steeper than the “summer”decline, W=17.01,
df =1, p<.001.
Moreover, student race significantly predicted
the latent “school year”slope, unstandardized
b=0.10, p=.036, 95% CI [0.18, 0.01]; model
fit: v
2
(26) =54.39, p<.001; RMSEA =.06, 90% CI
[0.04, 0.09]; CFI =.96, and not predict the summer
slope (p=.088). In sum, African American students
lost more trust during the school year than their
White peers. They did not lose more trust over the
summer.
Which school experiences might account for the
growth of the trust gap?
Procedural injustice in school discipline. Students
may discern whether to trust the school based on
the social reality of whether suspicion and punish-
ments primarily target members of their racial and
ethnic minority groups. Students’private grades
may matter as well (which is why we control for
them), but discipline-related events, such as a tea-
cher sending a student to the principal, are more
readily observable to all children.
Official records revealed that African American
students received significantly more discipline
throughout middle school. This was true at every
grade level in the continuous data and when
inspecting a dichotomous measure of whether a
student had any incident at all (sixth grade:
White =24%, African American =55%; seventh
grade: White =40%, African American =67%;
eighth grade: White =41%; African Ameri-
can =69%). Although disciplinary events rose with
time for both racial groups, the largest race gap
was apparent in sixth grade.
This racial group difference does not appear to
be driven by differences in social class or in aca-
demic achievement. In a Poisson regression predict-
ing number of discipline incidents in sixth to eighth
grades with student race and relevant covariates
(see above) a substantial student race gap in disci-
pline remained, unstandardized b=0.90 events, t
(269) =13.78, p<.001, 95% CI [0.77, 1.03], and this
was relatively undiminished compared with the
race gap with no covariates (raw gap: 5.8 incidents,
covariate-adjusted gap: 4.16 incidents). Thus, on the
basis of the allocation of punishment in this school,
African American students may have wondered
whether the rules of the institution were being
applied fairly to their group.
This is especially apparent when limiting analy-
ses to what we call “judgment call”incidents—op-
erationally, “defiance”and “disobedience,”as
coded by the school. In much prior research, there
has been no group disparity in discipline for more
objectively apparent infractions—such as bringing a
weapon to school—but African American students
were far more likely to be disciplined for subjective
infractions such as disrespect, disobedience, loiter-
ing, or excessive noise (Gregory & Weinstein, 2008;
Skiba et al., 2002). Students’levels of trust may be
more affected by subjective than objective incidents.
To be disciplined for an objectively apparent mat-
ter, like fighting, is difficult to quarrel with. But to
be disciplined for talking out of turn, and to see fel-
low African Americans more frequently sentenced
to such disciplinary measures, is perhaps more
likely to raise questions about teachers’trustworthi-
ness.
In the first cohort, the only cohort with school
codes for incident type, African American students
outnumbered their White peers roughly three to
one for defiance or disobedience (25% vs. 8%,
respectively), a significant difference v
2
(1) =6.52,
p=.012, and one that corresponds to the group dif-
ference in an analyses of all 17 middle schools in a
large district (Skiba et al., 2002). For objective inci-
dents such as cheating, fighting, or obscene ges-
tures, the race gap was not significant, (7% vs. 4%),
v
2
(1) =0.56, p=.453.
Awareness of bias in enforcement of school poli-
cies. In addition to receiving more discipline, Afri-
can American students also perceived bias in
disciplinary sentencing (see Figure 2). Analyses of
the awareness of bias survey items showed that
African American students were more likely to
expect that they, and not their White peers, would
be disciplined for the same events (e.g., being in the
hallway without a hall pass), at every measurement
occasion, ps<.05. White students, by contrast, saw
no bias in discipline. At every measurement occa-
sion except the last one, 80% of White students
expected equal treatment for White and African
American students (Figure 2). Only at the end of
eighth grade did White students show a significant
tendency to see bias against African Americans, but
they still saw less bias than African American stu-
dents did at the beginning of sixth grade, as shown
in Figure 2. By contrast, for African American stu-
dents, fewer than 55% expected equal treatment at
every wave except the first month of sixth grade.
666 Yeager, Purdie-Vaughns, Hooper, and Cohen
Recursive process. A parallel growth curve model
explored the relationship between African Ameri-
can students’awareness of bias in disciplinary deci-
sions and their level of trust. The model estimated
a latent intercept at fall of sixth grade as well as a
latent slope from that point until spring of eighth
grade, separately for both awareness of bias and
trust (Figure 3).
We found evidence consistent with the recursive
process displayed in Figure 1. African American
students who reported greater awareness of disci-
plinary bias (reverse-coded) in the fall of sixth
grade also reported lower trust at the same time
point (i.e., the intercepts were correlated), unstan-
dardized b=0.68, t(132) =3.97, p<.001, 95% CI
[0.28, 1.054]. Next, a change in awareness of bias
(reverse-coded) predicted a concurrent change in
trust (i.e., the slopes were correlated), b=1.79,
t(132) =7.58, p<.001, 95% CI [0.96, 2.61] (these
results were no different when estimating the
“school year”slopes only). These correlational data
do not isolate which of the two constructs, mistrust
or awareness of disciplinary bias, have temporal
causal precedence, but rather demonstrate their
pattern of reciprocal reinforcement through time.
Furthermore, if perceived bias and mistrust rein-
forced each other, then perceived bias should
increase with time. Indeed, the growth curve model
found a significant slope for awareness of bias (not
reverse-coded), unstandardized b=0.06, p=.002,
showing that African American adolescents
perceived increasingly high levels of bias through
middle school. This was in spite of the fact, noted
previously, that the objective size of the gap in
disciplinary sentencing, if anything, decreased with
time (although disciplinary events rose for all
students, regardless of race, over time). In sum,
students who were more aware of bias reported
lower trust, and students who became more aware
of bias lost more trust—and vice versa.
This process did not appear to be driven by stu-
dents who were the most likely to be disciplined.
Controlling for students’fifth-grade discipline inci-
dents did not change the significance levels of
either of these results. Moreover, there was no
interaction involving fifth-grade discipline (in a
multiple-group analysis), suggesting that the pat-
terns held for students with a history of disciplinary
problems and those without one. Furthermore, Afri-
can American students showed a significant decline
in trust (i.e., a significant slope) even when con-
fined to those who never received a disciplinary
infraction during middle school. These supplemen-
tal analyses dovetail with past research suggesting
that vicarious awareness of procedural injustice can
affect institutional trust, even when one has not
Figure 3. Recursive processes in awareness of bias and school trust among African American middle school students: a parallel growth
curve analysis in Study 1.
Note. Unstandardized coefficients. Covariates were gender, prior achievement (fifth-grade grade point average and fall sixth-grade test
scores), and cohort. “Awareness of bias”is coded so that higher values correspond to greater expectations of equal treatment (i.e., lower
awareness of bias), matching Figure 2, so that all correlations with trust were expected to be positive. ***p<.001.
Middle School Trust Gap 667
been the direct recipient of unfair treatment (Fagan
& Tyler, 2005).
Does a Loss of Trust Predict Future Discipline
Incidents?
What implications might the loss of school trust
have for future behavior? A series of regression
analyses found that lower trust each spring was
associated with disciplinary behavior the following
year. Poisson regressions predicting discipline inci-
dents showed that lower trust in the spring of sixth
and seventh grades predicted African American
students’discipline incidents in the subsequent
year: spring sixth-grade trust predicting seventh-
grade discipline, b=0.26, t(109) =5.53, p<.001,
95% CI [0.35, 0.17]; spring seventh-grade trust
predicting eighth grade discipline, b=0.22, t
(102) =4.172, p<.001, 95% CI [0.32, 0.12].
Hence, one year’s level of institutional trust pre-
dicted the next year’s level of behavioral defiance
of institutional policies.
Long-Term Consequences of Loss of Trust
Did a loss in trust predict long-term develop-
mental outcomes? We estimated residual scores
from regressions of spring trust on fall trust, for
each of the three school years: sixth, seventh, and
eighth. Lower values correspond to a greater loss of
trust. The advantage of the residual score is that it
is independent of either of the two scores used to
compute it (supplementary analyses using raw
change scores yielded the same substantive results).
Seventh-grade residual trust scores predicted on
time enrollment at a 4-year college for African
American students, standardized b=.19, Z=2.26,
p=.024, 95% CI [0.025, 0.36], meaning that stu-
dents who lost more trust than would be expected
in seventh grade were less likely to end up at a 4-
year college. (Similar results emerged in a supple-
mental ordinal regression analysis that examined
college enrollment at three levels, 0 =no college,
1=2-year college, 2 =4-year college, Z=2.38,
p=.018.) Fall eighth-grade absolute levels of trust
predicted college enrollment as well (p<.01), and
no other absolute or residual measure of trust did
(see Supporting Information).
Thus, it appears that the damage to African
American students’trust by the end of seventh
grade or the beginning of eighth grade significantly
predicted whether they made it to a 4-year college
the year after high school. Crucially, this was true
even after controlling for relevant predictors of
college enrollment such as premiddle school aca-
demic performance. In contrast, trust spring resid-
ual scores or within-year change scores in any year
were not associated with college enrollment for
White students (see Figure S5).
Effect of the “Wise Feedback”Intervention
One limitation with the analyses reported so far
is endogeneity. Students’social realities in their
institutions, their mental representations of them,
and their behaviors over time occur in a mutually
causal system, creating a confound for causal infer-
ence. We are fortunate, however, that it was possi-
ble to supplement a portion of the correlational
analysis with an experimental one. As noted, stu-
dents were randomly assigned at the individual
level to receive trust-sustaining “wise feedback”
from their seventh-grade teacher in the context of a
potentially unfair interpersonal event: being criti-
cized on a writing assignment (recall that, in a pre-
vious article reporting on these data, the note
sustained African American students’trust at
the end of seventh grade; Yeager et al., 2014,
Studies 1–2).
Eighth-grade discipline. Receiving a trust-sustain-
ing “wise feedback”intervention in the spring of
seventh grade significantly reduced African Ameri-
can students’eighth-grade discipline incidents, t
(43) =2.96, p=.005, d=.67. There was no effect
for White students (p=.75), and the Race 9Condi-
tion interaction was significant, t(81) =2.04,
p=.045. Figure 4 reports raw and covariate-
adjusted means. In the control condition, African
American students received four times more disci-
plinary citations as compared to White students. In
the treatment group that gap was halved. These
results represent a conceptual replication of
research that has changed teachers’chronic behav-
iors, improving students’feelings that teachers
respected them, thus reducing race-related disci-
pline disparities (Gregory et al., 2016; Okonofua,
Paunesku, et al., 2016).
These results support the theory presented in
Figure 1, which contends that adolescents engage
in inductive reasoning on the basis of interper-
sonal interactions but, after forming a mental
model, are also partly top–down in their assess-
ments of future attributionally ambiguous interac-
tions. Here, a trust-restoring experience with a
single teacher in seventh grade affected discipline
in eighth grade, even as students entered into new
interactions with different teachers and authority
figures in the school (for analogous year-over-year
668 Yeager, Purdie-Vaughns, Hooper, and Cohen
effects, see Gregory et al., 2016). This suggests that
it is possible to interrupt or at least slow the top-
down appraisals that contribute to worsening mis-
trust.
College enrollment. Next, the wise feedback inter-
vention in spring of seventh grade—the period
when we found that changes in trust were most
predictive of college outcomes—significantly
increased African American students’likelihood of
attending a 4-year college the year immediately fol-
lowing high school, 5.5 years posttreatment,
Z=2.58, p=.010 (Covariate-adjusted values: Con-
trol =40%, Wise feedback =70%, see Figure 4 for
raw values). Gender, prior achievement, and prior
trust all significantly predicted college enrollment,
but cohort did not; removing the latter did not
change the significance of the treatment effect
(p=.01). Treatment effects for White students were
nonsignificant (p<.5).
There were no treatment effects on whether ado-
lescents enrolled at any college at all (2 or 4 years),
Z=0.16, p=.880, only on 4-year college enroll-
ment. In a supplemental ordinal regression analysis
when 0 =no college enrollment, 1 =2-year college,
and 2 =4-year college, the treatment effect was
marginally significant, Z=1.69, p=.092.
Recall, however, that this experiment was con-
ducted among the subset of “B”and “C”average
students who were already more likely to attend
college than students with lower grades, and so
there was range restriction. Within the subset of
those who attended any college at all (N=37 of
Figure 4. Longitudinal effects of trust-restoring “wise feedback”intervention (administered spring of seventh grade) on eighth-grade
discipline incidents and on time enrollment in a 4-year college 5½years postintervention, in Study 1.
Note. Covariates include prior achievement (grade point average and test scores), preintervention trust, gender, and cohort. Covariate-
adjusted values for each racial group estimated at the mean for each racial group.
Middle School Trust Gap 669
44), we analyzed treatment effects on 4-year college
enrollment (1 =4-year college, 0 =2-year college,
NA =not enrolled in college). In that analysis,
effects of the wise feedback treatment were stron-
ger, Z=2.95, p=.003, with covariates. Thus, at
least in our small sample of middle-class “B”and
“C”African American students who eventually
attended college, the trust-sustaining “wise-feed-
back”intervention resulted in greater likelihood of
attending a 4-year college instead of a 2-year col-
lege.
Study 2
Method
Participants and Procedures
Study 2 was a 1-year cohort sequential study
that followed the entire middle school (sixth, sev-
enth, and eighth grade) at a rural school in Color-
ado for 1 year (N=206). Students were either
Hispanic or Latino (44%) or White (56%). Unlike
Study 1, the negatively-stereotyped ethnic minority
students were largely working class. These differ-
ences between the studies permitted us to assess
the generality of our findings to a new negatively
stereotyped group (Latino Americans) that was
relatively economically disadvantaged and living in
an altogether different geographic location from the
students in Study 1.
Survey procedures were analogous to Study 1.
Adolescents completed fall and spring survey mea-
sures assessing school trust and awareness of proce-
dural injustice, identical to Study 1, except the
procedural injustice in school policies survey mea-
sures (after piloting) referred to “Hispanic”stu-
dents, not “African American”students. Students
were surveyed only for 1 year of middle school,
and because this study was conducted years after
Study 1, students have not yet been tracked
through college. Furthermore, discipline incidents
in the year after the study were not available.
Results
Emergence of the Latino–White Trust Gap in Middle
School
As in Study 1, a significant (p<.05) ethnic group
gap in school trust emerged during fall of 7th grade
and was sustained through the end of middle
school (see Figure 2); once again, the trust gap was
not significant in 6th grade. As in Study 1, the
largest loss in trust appeared in the 7th grade
school year, was strongest for minority rather than
majority students, with the largest gap in trust
between the two ethnic groups occurring in the
spring of 7th grade. Thus, analyses of school trust
in a cohort-sequential design with Latinos repli-
cated Study 1’s results with African Americans.
What Caused the Trust Gap?
Unlike Study 1, in the fall of sixth grade there
was no initial significant Latino–white difference in
expectations of bias in application of school policies
(Figure 2). However, by the spring of sixth grade a
significant difference emerged, and from then on
the results of Study 2 paralleled Study 1.
Next, we examined the relationship between
awareness of bias and changes in trust. Because
Study 2 is a cohort sequential design, not a true
longitudinal design like Study 1, the latent change
score model is not identified, as it requires at least
three observations. Therefore, we estimated cross-
lagged autoregressive models with Latino partici-
pants from all three grades (sixth, seventh, and
eighth) and tested the effects of fall awareness of
bias on spring trust.
The cross-lagged model showed that the extent
to which Latino students believed there was injus-
tice in school discipline in the fall was a significant
predictor of their subsequent trust of school in the
spring, in cross-lagged models (b=0.53, p<.001;
see Figure 5). The reverse was not true, suggesting
a path from awareness of bias to trust, and not vice
versa. This path was not significantly different for
any of the three grade levels, when constraining
paths across grade levels in a multigroup analysis
(see Supporting Information). (We have also recon-
ducted Study 1’s analyses using cross-lagged mod-
els, and they are reported in the online supplement,
Figure S6. These yield the same conclusions, with
the exception that both paths, from trust to bias
Figure 5. Cross-lagged models show that awareness of bias in
discipline predicts subsequent lower levels of trust among Latinos
in Study 2. Unstandardized regression coefficients. ***p<.001.
670 Yeager, Purdie-Vaughns, Hooper, and Cohen
and bias to trust, are significant at different waves.)
These findings extend the parallel growth curve
analysis in Study 1 by showing that awareness of
bias in the fall predicts lower trust of school in the
spring.
Discussion
In a Pew Center survey of adults in the United
States, 61% of African Americans and 53% of Lati-
nos reported low levels of trust in the fairness of
American society, as compared to only 32% of
White Americans (Taylor, Funk, & Clark, 2007).
Analogous gaps exist with regard to specific U.S.
institutions. For instance, 70% of African Americans
recently reported low levels of trust in the police
compared to 43% of White Americans (Jones, 2015).
Such racial and ethnic gaps in adults’levels of trust
are large. They may appear unavoidable, norma-
tive, and unsurprising. A key contribution of this
article is to offer a testable social and developmen-
tal psychological process model for how racial and
ethnic gaps in institutional trust emerge during
adolescence and affect long-term developmental
outcomes (Figure 1). A second contribution is to
show that such gaps and their consequences need
not be inevitable.
Specifically, this research used a theory-informed
developmental frame to understand the formation
of institutional trust among racial and ethnic minor-
ity adolescents and to explore its behavioral impli-
cations over an 8-year period. Trust declined every
semester of middle school, and this decline hap-
pened more quickly and strongly for African Amer-
icans (Study 1) and Latino (Study 2) students.
Racial and ethnic minority students’decline in trust
was predicted by their awareness of bias—that is,
their awareness of the possibility of procedural
injustice at their school. This finding was confirmed
using a parallel growth curve model (Study 1) and
a cross-lagged model (Study 2). These results are
consistent with much prior research showing that
trust is tied to perceptions of procedural justice and
predicts institutional compliance (Levi, 1997; Tyler
& Blader, 2003).
One intriguing aspect of the process shown in
Figure 1 is its recursive nature. Once students’
sense of trust or distrust was formed, it seemed to
feed off its consequences, producing perceptions of
procedural injustice that caused trust to decline fur-
ther (also see Fagan & Tyler, 2005; Gregory &
Weinstein, 2008). Moreover, that decline in trust
seemed to increase the likelihood of discipline
infractions, creating the very social reality that pre-
cipitated it. These feedback loops proceed often hid-
den from the view of teachers and administrators
because they unfold slowly and are partly psycho-
logical in nature. But their cumulative effect is a
large trust gap by seventh grade that disfavored
racial and ethnic minority students. Years later, the
drop in trust in the transition to seventh grade and
then eighth grade seemed to have lingering conse-
quences, in the form of lower 4-year college enroll-
ment for African Americans.
Trust, it seems, sat “in the middle”between
social reality and later behavioral outcomes such as
disciplinary infractions and college enrollment (cf.
Olson & Dweck, 2008). We know this from an inter-
vention in Study 1 that experimentally bolstered
African Americans’sense of trust in the face of
sharp criticism of their work in the seventh grade
(previously shown in Yeager et al., 2014, Study 2).
Because the link between trust and later outcomes
depends on a continual feedback loop, an early
experience that refuted the plausibility of procedu-
ral injustice had long-term effects, presumably
through a kind of developmental cascade from trust
to engagement and into educational pathways (cf.
Masten et al., 2005).
African American seventh graders who received
wise feedback on an essay—conveying that the tea-
cher believed in their potential to reach a higher
standard, thus reassuring students that they would
be seen based on their merits rather than through
the lens of a negative stereotype about the intellec-
tual ability of their racial group—benefited. Our
intervention was based on a body of previous theo-
retical and empirical research (Cohen et al., 1999;
Gregory & Weinstein, 2008; Shouse, 1996) and was
timed to the emergence of the trust gap (see Cohen,
Garcia, Purdie-Vaughns, Apfel, & Brzustoski, 2009).
Although the objective experience of receiving
“wise feedback”was short, the psychological and
developmental consequences seemed long-lasting.
Adolescents receiving the note were assigned fewer
disciplinary infractions later according to official
records and, nearly 6 years later, were more likely
to attend a 4-year college according to the NSC.
Trust, and the trust-restoring wise feedback
intervention, did not predict subsequent college
persistence for White students (Figure S5). Why
not? For students with group-based advantages,
such as majority-group students, there may simply
be greater redundancy in the system. When one’s
group is positively stereotyped and overrepresented
in a domain, a loss of trust or a poor relationship
with a teacher might be a temporary setback but
Middle School Trust Gap 671
not fatal. For a stigmatized student, however, out-
comes might depend more on the accident of hav-
ing a teacher who believes in them (for an
analogous example from low-income children in
early childhood, see Tucker-Drob, 2012). Thus,
although the present results offer an exciting possi-
bility for what might be done to prevent race-based
achievement gaps, they simultaneously act as an
indictment of the educational system, by demon-
strating that minority students’outcomes are more
contingent on everyday experiences of respect—ex-
periences that may occur too rarely and by happen-
stance.
Limitations and Future Directions
The wise feedback experiment has limitations.
First, the evidence is based on only one school. Sec-
ond, the sample size was small, and small sample
sizes can lead to false-positive findings or to over-
stated effect sizes when researchers have flexibility
in data analysis. Here, sample size was constrained
by the number of B and C students in the two
cohorts, and analysis was constrained to be parallel
across types of analysis—correlational and experi-
mental—which yielded similar results. In addition,
the results are robust across many (but not all) alter-
native model specifications (see both significant and
nonsignificant model specifications at osf.io/3hpu8).
At the same time, the overall sample size was
approximately 70% of the longitudinal sample for
the Perry Preschool project, which has had a
tremendous influence on early childhood policy
(e.g., Heckman, 2006). Furthermore, unlike the
Perry Preschool project, in this study there were no
problems with randomization or missing data (be-
cause random assignment was effective, Yeager
et al., 2014; and outcome data were collected from
the NSC). We also note that prior psychological
interventions that initially involved a few dozen
treated students (e.g., Aronson, Fried, & Good,
2002) later showed reproducible findings when con-
ducted with many thousands of students (Paunesku
et al., 2015) and when analyses were preregistered
and data were collected by third-party research
firms (Yeager, Romero, et al., 2016).
Nevertheless, it will be important for future
experiments to reproduce our method with a larger
sample and in more heterogeneous contexts. Doing
this will not be without challenges. The present
experiments require working closely with schools
over several years to guide teachers in designing
assignments and hand-writing personal notes to
students, and tracking these students over the
ensuing 6 years. Moreover, we know very little
about what contextual conditions might moderate
the effects of wise feedback, but we imagine they
are plentiful.
Although we have focused on the middle school
transition, the processes documented here did not
begin in sixth grade and do not stop in eighth
grade. It will be critical in future research to under-
stand how to prevent experiences of racial and eth-
nic discrimination that begin early in childhood and
also to develop methods to regain racial and ethnic
minority students’trust after they transition to high
school. For instance, one possibility is that students
reengage their potential to trust when they transi-
tion to a novel institution—asking themselves
whether the “old codes”apply or not (cf. Coates,
2015). Perhaps timely intervention at later life tran-
sitions might approximate the effects of the wise
feedback intervention documented here (for an
example among college students, see Yeager, Wal-
ton, et al., 2016).
We emphasize that our theoretical model does
not “psychologize”the social problem of racial dis-
parities in discipline and college enrollment. Nor
does it “sociologize”it and ascribe outcomes solely
to unalterable structural causes. It is, as we show in
Figure 1, the interaction of structure with psyche
that drives outcomes over time (Cohen, Garcia, &
Goyer, 2017). School presents a social reality to
which all students, especially negatively stereo-
typed minority groups, must psychologically adapt.
That psychological adaptation, in turn, can reinforce
the social reality, as when African American middle
school students act out against a system they per-
ceive as unjust. It is neither the attributes of the
child, such as a troublesome nature or a behavioral
disorder, nor the social environment alone that is
the driver of inequality. Rather it is the unfolding
transaction between the child and the environment
(Cohen et al., 2017; Gregory & Weinstein, 2008;
Okonofua, Walton, et al., 2016).
Our approach echoes Bronfenbrenner’s (1977)
early formulation of ecological systems theory and
aligns with contemporary social-cognitive theories
of development (Olson & Dweck, 2008). Mental
representations, the product of observing and inter-
acting with a social context, continue to exert an
influence on development for years to come. Yet by
the logic of causal mediation, intervening in a
timely way slowed the psychological accretions of
injustice and lessened its lasting harm.
Our theoretical perspective is consistent with
phenomenological variants on ecological systems
theories, which address the social, structural,
672 Yeager, Purdie-Vaughns, Hooper, and Cohen
cultural, and historical contexts in which youth
develop (Spencer, 2006). A mental model of trust
can act as a protective factor or a risk factor. Trust
allows a person to reap the rewards of engagement
with the system, but it also puts them at risk of
being taken advantage of. Trust, absent justice,
might be misplaced, but well-placed trust can help
people thrive. Hence, institutions would do well to
simultaneously improve fairness and reduce bias
while also addressing the psychological legacy of
previous injustices. Said differently, wise feedback
is not, on its own, the solution to race gaps in disci-
pline. Schools and teachers need to be trustworthy.
When they are, then wise feedback can help clear
the attributional air and redirect students’working
mental models of the institution, so that students
can profit from the relationships and instruction in
their schools.
These data inform education policy and practice.
By experimentally testing the wise feedback inter-
vention, it demonstrates a method for helping teach-
ers create a classroom climate that is more likely to
maintain the trust of students who may contend with
discrimination (also see Gregory et al., 2016; Okono-
fua, Paunesku, et al., 2016). Based on prior correla-
tional (Gregory & Weinstein, 2008; Shouse, 1996) and
experimental research (Cohen & Steele, 2002; Cohen
et al., 2009; Yeager et al., 2014), the intervention was
not only precisely targeted to address the key psy-
chological process but was timed to intervene on that
process at a crucial stage: when mistrust was likely
to emerge and to exert a growing impact. This is
because a process can be easier to affect at its begin-
ning. Once it has accumulated consequences—the
child is labeled as a “troublemaker,”sees his perma-
nent record tainted by a suspension—the process
will have a momentum that is much harder to halt
(Okonofua, Paunesku, et al., 2016; Okonofua, Wal-
ton, et al., 2016).
Of course, truly “wise”educators do not simply
append notes to essays and end their interventions
there. Instead, they continually send the message
that their students are capable, valued, and
respected, weaving it into the culture of the class-
room. Our studies demonstrate the way that larger
cultural forces infuse the interactions between tea-
chers and students. How to break free of their influ-
ence is a craft that requires both wisdom and tact.
References
Aronson, J. M., Fried, C. B., & Good, C. (2002). Reducing
the effects of stereotype threat on African American
college students by shaping theories of intelligence.
Journal of Experimental Social Psychology,38, 113–125.
doi:10.1006/jesp.2001.1491
Baumrind, D. (1997). The discipline encounter: Contem-
porary issues. Aggression and Violent Behavior,2, 321–
335. doi:10.1016/S1359-1789(97)00018-9
Bigler, R. S., Averhart, C. J., & Liben, L. S. (2003). Race
and the workforce: Occupational status, aspirations, and
stereotyping among African American children. Develop-
mental Psychology,39, 572–580. doi:10.1037/0012-1649.39.3.
572
Bigler, R. S., & Liben, L. S. (2006). A developmental inter-
group theory of social stereotypes and prejudice.
Advances in Child Development and Behavior,34,39–89.
doi:10.1016/S0065-2407(06)80004-2
Bronfenbrenner, U. (1977). Toward an experimental ecol-
ogy of human development. American Psychologist,32,
513–531. doi:10.1037/0003-066X.32.7.513
Brown, C. S., & Bigler, R. S. (2005). Children’s perceptions
of discrimination: A developmental model. Child Develop-
ment,76, 533–553. doi:10.1111/j.1467-8624.2005.00862.x
Bryk, A. S., & Schneider, B. (2002). Trust in schools: A core
resource for improvement. New York, NY: Russell Sage
Foundation.
Coates, T.-N. (2015). Between the world and me. New York,
NY: Random House Publishing Group.
Cohen, G. L., & Garcia, J. (2005). “Iamus”: Negative
stereotypes as collective threats. Journal of Personality
and Social Psychology,89, 566–582. doi:10.1037/0022-
3514.89.4.566
Cohen, G. L., Garcia, J., & Goyer, J. P. (in press). Turning
point: Targeted, tailored, and timely psychological
intervention. In A. Elliot, C. S. Dweck & D. S. Yeager
(Eds.) Handbook of competence and motivation (2nd ed.):
Theory and application. New York, NY: Guilford.
Cohen, G. L., Garcia, J., Purdie-Vaughns, V., Apfel, N., &
Brzustoski, P. (2009). Recursive processes in self-affir-
mation: Intervening to close the minority achievement
gap. Science,324, 400–403. doi:10.1126/science.1170769
Cohen, G. L., & Steele, C. M. (2002). A barrier of mistrust:
How negative stereotypes affect cross-race mentoring.
In J. M. Aronson (Ed.), Improving academic achievement:
Impact of psychological factors on education (pp. 303–327).
Bingley, UK: Emerald Group.
Cohen, G. L., Steele, C. M., & Ross, L. D. (1999). The
mentor’s dilemma: Providing critical feedback across
the racial divide. Personality and Social Psychology Bul-
letin,25, 1302–1318. doi:10.1177/0146167299258011
Coopersmith, J. (2009). Characteristics of public, private, and
bureau of indian education elementary and secondary school
teachers in the United States: Results from the 2007–08 schools
and staffing survey. First Look. NCES 2009-324. Washing-
ton D.C.: National Center for Education Statistics.
Dynarski, S. M., Hemelt, S. W., & Hyman, J. M. (2015).
The missing manual: Using National Student Clearing-
house data to track postsecondary outcomes. Educa-
tional Evaluation and Policy Analysis,37, 53s–79s. doi:10.
3102/0162373715576078
Middle School Trust Gap 673
Fagan, J., & Tyler, T. R. (2005). Legal socialization of chil-
dren and adolescents. Social Justice Research,18, 217–
241. doi:10.1007/s11211-005-6823-3
Gambetta, D. (1988). Can we trust trust? In D. Gambetta
(Ed.), Trust: Making and breaking cooperative relations (pp.
213–235). New York, NY: Basil Blackwell.
Goffman, E. (1963). Stigma: Notes on the management of
spoiled identity. New York, NY: Simon & Schuster.
Gregory, A., Hafen, C. A., Ruzek, E., Mikami, A. Y.,
Allen, J. P., & Pianta, R. C. (2016). Closing the racial
discipline gap in classrooms by changing teacher practice.
School Psychology Review,45,171–191.
Gregory, A., & Weinstein, R. S. (2004). Connection and
regulation at home and in school predicting growth in
achievement for adolescents. Journal of Adolescent
Research,19, 405–427. doi:10.1177/0743558403258859
Gregory, A., & Weinstein, R. S. (2008). The discipline gap
and African Americans: Defiance or cooperation in the
high school classroom. Journal of School Psychology,46,
455–475. doi:10.1016/j.jsp.2007.09.001
Harber, K. D. (1998). Feedback to minorities: Evidence of
a positive bias. Journal of Personality and Social Psychol-
ogy,74, 622–628. doi:10.1037/0022-3514.74.3.622
Heckman, J. J. (2006). Skill formation and the economics
of investing in disadvantaged children. Science,312,
1900–1902. doi:10.1126/science.1128898
Irvine, J. J. (2003). Educating teachers for diversity: Seeing
with a cultural eye. New York, NY: Teachers College
Press.
Jones, J. M. (2015, June 19). In U.S., confidence in police
lowest in 22 years. Omaha, NE: Gallup. Retrieved from
http://www.gallup.com/poll/183704/confidence-police-
lowest-years.aspx
Killen, M., Mulvey, K. L., & Hitti, A. (2013). Social exclu-
sion in childhood: A developmental intergroup per-
spective. Child Development,84, 772–790. doi:10.1111/
cdev.12012
Laupa, M., Turiel, E., & Cowan, P. A. (1995). Obedience
to authority in children and adults. In M. Killen & D.
Hart (Eds.), Morality in everyday life: Developmental per-
spectives (pp. 131–165). Cambridge, UK: Cambridge
University Press.
Lee, K. H., Siegle, G. J., Dahl, R. E., Hooley, J. M., & Silk,
J. S. (2014). Neural responses to maternal criticism in
healthy youth. Social Cognitive and Affective Neuroscience,
10, 902–912. doi:10.1093/scan/nsu133
Lepper, M. R., & Woolverton, M. (2002). The wisdom
of practice: Lessons learned from the study of
highly effective tutors. In J. M. Aronson (Ed.), Improv-
ing academic achievement: Impact of psychological factors
on education (pp. 135–158). San Diego, CA: Academic
Press.
Levi, M. (1997). Consent, dissent, and patriotism. New York,
NY: Cambridge University Press.
Levi, M., Sacks, A., & Tyler, T. (2009). Conceptualizing
legitimacy, measuring legitimating beliefs. American
Behavioral Scientist,53, 354–375. doi:10.1177/
0002764209338797
Lewin, K., Lippitt, R., & White, R. K. (1939). Patterns of
aggressive behavior in experimentally created social cli-
mates. The Journal of Social Psychology,10, 269–299.
doi:10.1080/00224545.1939.9713366
Lord, C. G., Ross, L., & Lepper, M. R. (1979). Biased
assimilation and attitude polarization: The effects of
prior theories on subsequently considered evidence.
Journal of Personality and Social Psychology,37, 2098–
2109. doi:10.1037/0022-3514.37.11.2098
Masten, A. S., Roisman, G. I., Long, J. D., Burt, K. B.,
Obradovi
c, J., Riley, J. R., . . . Tellegen, A. (2005).
Developmental cascades: Linking academic achieve-
ment and externalizing and internalizing symptoms
over 20 years. Developmental Psychology,41, 733–746.
doi:10.1037/0012-1649.41.5.733
McKown, C. (2013). Social equity theory and racial-ethnic
achievement gaps. Child Development,84, 1120–1136.
doi:10.1111/cdev.12033
McKown, C., & Weinstein, R. S. (2003). The development
and consequences of stereotype consciousness in mid-
dle childhood. Child Development,74, 498–515.
Muth
en, L. K., & Muth
en, B. O. (1998-2011). Mplus user’s
guide. Sixth Edition. Los Angeles, CA: Author.
Okonofua, J. A., & Eberhardt, J. L. (2015). Two strikes:
Race and the disciplining of young students. Psychologi-
cal Science,26, 617–624. doi:10.1177/0956797615570365
Okonofua, J. A., Paunesku, D., & Walton, G. M. (2016). A
brief intervention to encourage empathic discipline cuts
suspension rates in half among adolescents. Proceedings
of the National Academy of Sciences of the United States of
America,113, 5221–5226.
Okonofua, J. A., Walton, G. M., & Eberhardt, J. L. (2016).
A vicious cycle: A social-psychological account of
extreme racial disparities in school discipline. Perspec-
tives on Psychological Science,11, 381–398.
Olson, K. R., & Dweck, C. S. (2008). A blueprint for
social cognitive development. Perspectives on Psychologi-
cal Science,3, 193–202. doi:10.1111/j.1745-6916.2006.
00007.x
Paunesku, D., Walton, G. M., Romero, C., Smith, E. N.,
Yeager, D. S., & Dweck, C. S. (2015). Mindset interven-
tions are a scalable treatment for academic under-
achievement. Psychological Science,26, 284–293. doi:10.
1177/0956797615571017
Peper, J. S., & Dahl, R. E. (2013). The teenage brain: Surg-
ing hormones—brain-behavior interactions during
puberty. Current Directions in Psychological Science,22,
134–139.
President’s Task Force on 21st Century Policing. (2015).
Final report of The President’s Task Force on 21st Century
Policing. Washington, DC: Office of Community Ori-
ented Policing Services.
Purdie-Vaughns, V., Steele, C. M., Davies, P. G., Ditl-
mann, R., & Crosby, J. R. (2008). Social identity contin-
gencies: How diversity cues signal threat or safety for
African Americans in mainstream institutions. Journal of
Personality and Social Psychology,94, 615–630. doi:10.
1037/0022-3514.94.4.615
674 Yeager, Purdie-Vaughns, Hooper, and Cohen
Roberts, B. W., & Caspi, A. (2003). The cumulative conti-
nuity model of personality development: Striking a bal-
ance between continuity and change in personality
traits across the life course. In U. M. Staudinger & U.
E. R. Lindenberger (Eds.), Understanding human
development: Lifespan psychology in exchange with other
disciplines (pp. 183–214). Dordrecht, NL: Kluwer
Academic.
Ruble, D. N., Alvarez, J., Bachman, M., Cameron, J.,
Fuligni, A., Garcia Coll, C., & Rhee, E. (2004).
The development of a sense of “we”: The emergence
and implications of children’s collective identity. In
M. Bennett & F. Sani (Eds.), The development of the
social self (pp. 29–76). New York, NY: Psychology
Press.
Sagar, H. A., & Schofield, J. W. (1980). Racial and behav-
ioral cues in black and white children’s perceptions of
ambiguously aggressive acts. Journal of Personality and
Social Psychology,39, 590–598.
Sellers, R. M., Copeland-Linder, N., Martin, P. P., &
Lewis, R. L. (2006). Racial identity matters: The rela-
tionship between racial discrimination and psychologi-
cal functioning in African American adolescents. Journal
of Research on Adolescence,16, 187–216.
Shouse, R. C. (1996). Academic press and sense of com-
munity: Conflict, congruence, and implications for stu-
dent achievement. Social Psychology of Education,1(1),
47–68.
Skiba, R. J., Michael, R. S., Nardo, A. C., & Peterson, R.
L. (2002). The color of discipline: Sources of racial and
gender disproportionality in school punishment. The
Urban Review,34, 317–342. doi:10.1023/A:
1021320817372
Spencer, M. B. (2006). Phenomenology and ecological sys-
tems theory: Development of diverse groups. In W.
Damon & R. M. Lerner (Eds.), Handbook of child psychol-
ogy, Vol. 1: Theoretical models of human development (6th
ed., pp. 829–893). New York, NY: Wiley.
Swanson, D. P., Cunningham, M., Youngblood, J., &
Spencer, M. B. (2009). Racial identity development dur-
ing childhood. In H. A. Neville, B. M. Tynes, & S. O.
Utsey (Eds.), Handbook of African American psychology
(pp. 269–281). Thousand Oaks, CA: Sage.
Taylor, P., Funk, C., & Clark, A. (2007). Americans and
social trust: Who, where and why. Washington, DC: Pew
Research Center.
Terburg, D., & van Honk, J. (2013). Approach–avoidance
versus dominance–submissiveness: A multilevel neural
framework on how testosterone promotes social status.
Emotion Review,5, 296–302.
Treisman, U. (1992). Studying students studying calculus:
A look at the lives of minority mathematics students in
college. The College Mathematics Journal,23, 362–372.
doi:10.2307/2686410
Tucker-Drob, E. M. (2012). Preschools reduce early aca-
demic-achievement gaps: A longitudinal twin
approach. Psychological Science,23, 310–319. doi:10.
1177/0956797611426728
Tyler, T. R. (2006). Restorative justice and procedural jus-
tice: Dealing with rule breaking. Journal of Social Issues,
62, 307–326. doi:10.1111/j.1540-4560.2006.00452.x
Tyler, T. R., & Blader, S. L. (2003). The group engagement
model: Procedural justice, social identity, and coopera-
tive behavior. Personality and Social Psychology Review,
7, 349–361. doi:10.1207/S15327957PSPR0704_07
Tyler, T. R., Fagan, J., & Geller, A. (2014). Street stops
and police legitimacy: Teachable moments in young
urban men’s legal socialization. Journal of Empirical
Legal Studies,11, 751–785. doi:10.1111/jels.12055
Tyler, T. R., Goff, P. A., & MacCoun, R. J. (2015). The
impact of psychological science on policing in the uni-
ted states procedural justice, legitimacy, and effective
law enforcement. Psychological Science in the Public Inter-
est,16,75–109. doi:10.1177/1529100615617791
Vanderbilt, K. E., Liu, D., & Heyman, G. D. (2011). The
development of distrust. Child Development,82, 1372–
1380. doi:10.1111/j.1467-8624.2011.01629.x
Vasquez, J. A. (1988). Contexts of learning for minority
students. The Educational Forum,52, 243–253. doi:10.
1080/00131728809335490
Yeager, D. S., Purdie-Vaughns, V., Garcia, J., Apfel, N.,
Brzustoski, P., Master, A., . . . Cohen, G. L. (2014).
Breaking the cycle of mistrust: Wise interventions to
provide critical feedback across the racial divide. Jour-
nal of Experimental Psychology: General,143, 804–824.
doi:10.1037/a0033906
Yeager, D. S., Romero, C., Paunesku, D., Hulleman, C. S.,
Schneider, B., Hinojosa, C., . . . Dweck, C. S. (2016).
Using design thinking to improve psychological inter-
ventions: The case of the growth mindset during the
transition to high school. Journal of Educational Psychol-
ogy,108, 374–391. doi:10.1037/edu0000098
Yeager, D. S., & Walton, G. M. (2011). Social-psychologi-
cal interventions in education: They’re not magic.
Review of Educational Research,81, 267–301. doi:10.3102/
0034654311405999
Yeager, D. S., Walton, G. M., Brady, S. T., Akcinar, E. N.,
Paunesku, D., Keane, L., . . . Dweck, C. S. (2016).
Teaching a lay theory before college narrows achieve-
ment gaps at scale. Proceedings of the National Academy
of Sciences of the United States of America,113, E3341–
E3348. doi:10.1073/pnas.1524360113
Supporting Information
Additional supporting information may be found in
the online version of this article at the publisher’s
website:
Figure S1. Exploratory Mediation Analysis of
Effects of Seventh-Grade Trust (Measured, Top
Panel; or Manipulated, Bottom Panel) on on time 4-
Year College Enrollment, Via Eighth-Grade Behav-
iors, for African American Participants Only
Figure S2. Cross-Lagged Models Show That
Awareness of Bias Predicts Subsequent Trust for
Middle School Trust Gap 675
Seventh- and Eighth-Grade Hispanic/Latino Stu-
dents, Controlling for Gender
Figure S3. Individual Differences in Awareness
of Anti-African American Procedural Injustice in
School Policies Do Not Predict the Onset of Overall
Mental Models of Trust of School Among White
Middle School Students in Study 1
Figure S4. The “Trust Residual Predicts College
Enrollment”Model, Excluding Treated Students
Figure S5. The “Trust Residual Predicts College
Enrollment”Model, White Students Only
Figure S6. Cross-Lagged Analyses Among
African American Participants in Study 1, Showing
the Reciprocal Relations Between Perceived Bias
and Loss of Trust Across the Middle School Years
Table S1. Study 1 Racial Group Differences in
School Trust
Table S2. Study 2 Ethnic Group Differences in
School Trust
Table S3. Descriptive Statistics for Study Vari-
ables (Study 1)
Table S4. Descriptive Statistics for Study Vari-
ables (Study 2)
676 Yeager, Purdie-Vaughns, Hooper, and Cohen