ArticlePDF Available

Abstract and Figures

This chapter provides a brief overview of research on stereotype threat, and considers whether this phenomenon is specific to minority groups (defined as low status groups), or whether similar deficits may also be observed in groups that generally enjoy a high status in society but that are negatively stereotyped in a specific domain. We then review a number of individual difference variables that moderate stereotype threat and that may explain why some people are highly vulnerable to stereotype activation while others appear to resist its influence. Next, we consider what processes drive stereotype threat, including anxiety, intrusive thoughts, shift towards caution, expectancy, and disengagement. In the subsequent section we compare the stereotype threat model with other theories dealing with the link between stereotypes and performance, in particular self-fulfilling prophecy and the expectancy value model. The final sections of the chapter concern areas of application in which stereotype threat may account for performance gaps between social groups, and how to prevent it.
Content may be subject to copyright.
Stereotype threat: When minority members
underperform
Anne Maass and Mara Cadinu
University of Padova, Italy
In 1954 Allport discussed the pernicious effects of prejudice on the members
of stigmatised groups. According to Allport, one of the most insidious of
these effects occurs when a member of a stigmatised group confirms with his/
her behaviour the negative stereotypes prevalent in the society, thus
‘‘reassuring’’ the prejudiced majority members that their negative stereo-
types are indeed justified. Despite this early interest, minority behaviour has
subsequently been typically underinvestigated in favour of an extensive
effort to investigate attitudes and behaviours held by majority members.
However, recent work by Crocker and Major (1989), Steele and Aronson
(1995), Swim and Stangor (1998) and others has contributed to a renewed
interest in the so-called minority perspective, in an attempt to better
understand the effects of stereotypes and prejudice on minority members.
One important contribution in this direction has been the recent model of
stereotype threat proposed by Steele and collaborators (Spencer, Steele, &
Quinn,1999; Steele & Aronson, 1995). These authors claim that negative
stereotypes are in part responsible for the underperformance of minority
members in stereotype-relevant domains. More specifically, those tasks for
which a negative association exists between the task domain and the
minority group will represent a threat for minority members; their
preoccupation with inadvertently confirming the stereotype will in turn
lead to a decrease in performance. For example, women are expected to
perform poorly on maths tests if the gender stereotype ‘‘Women are poor at
maths’’ becomes salient in the test situation.
#
2003 Psychology Press Ltd
http://www.tandf.co.uk/journals/pp/02699931.html DOI: 10.1080/10463280340000072
1
Address correspondence to Anne Maass, DPSS, University of Padova, Via Venezia, 8,
35139 Padova, Italy. E-mail: anne.maass@unipd.it
We would like to thank Francesco Foroni for his great help in conducting a pilot study in
the USA.
PERS0017
EUROPEAN REVIEW OF SOCIAL PSYCHOLOGY,
2003, 0, 000–000
In this chapter, we will first provide a brief overview of stereotype threat
research, with particular attention to the different research paradigms used
to test the model. We will then ask whether stereotype threat is specific to
minority groups, defined as low-status groups, or whether similar deficits
may also be observed in groups that generally enjoy a high statu s in society
but that are negatively stereotyped in a specific domain (e.g., men with
reference to social intelligence). Subsequ ently, we will review a number of
individual difference variables that moderate stereotype threat and that may
explain why some people are highly vulnerable to stereotype activation while
others appear to resist its influence. We will then ask when and how
stereotype threat related deficits may develop during childhood. Subse-
quently, we will address the question of what processes are driving
stereotype threat, an issue that continues to be the prime challenge to
researchers; in this section, we will consider a number of hypotheses
including anxiety, intrusive thoughts, shift towards caution, expecta ncy, and
disengagement. In the subsequent section we will compare the stereotype
threat model with other theories dealing with the link between stereotypes
and performance, in particular self-fulfilling prophecy and the expectancy
value model. We will ask whether stereotype threat does indeed represent a
distinct model or whether it is largely redundant with respect to previous
theories. The final sections will be dedicated to the areas of application in
which stereotype threat may account for performance gaps between social
groups, and to ways of preventing stereotype threat. Throughout this
chapter, we will draw both on the general literature and on our own studies,
including c urrent work. Unless mentioned otherwise, we will use the term
‘‘minority’’ as referring to a social group that has a relatively low status in
society, even if it represents a numerical majority (e.g., women), and that is
negatively stereotyped in a specific area of competence (e.g., women in
mathematics; Blacks in verbal intelligence; low social class in academics;
elderly in memory tasks).
RESEARCH PARADIGMS
In the paradigm originally proposed by Steele and Aronson (1995), minority
and majority members were asked to solve a task associated with a negative
minority stereotype (e.g., women on maths tasks, Blacks on verbal tasks). In
the stereotype threat condition the test was described as diagnostic of the
participants’ performance, whereas in the control condition it was described
as a simple exercise. The typical finding emerging from these studies is that
minority members (women, Blacks) show significantly poorer performance
when they believe that the test is diagnostic of their abilities than in the
control condition. On the contrary, the experimental manipulation does not
affect majority members’ performance (men, Whites). According to Steele
2
MAASS AND CADINU
and Aronson, the decline in performance can be attributed to stereotype
threat: The diagnostic test will activate the negative stereotype about the
ingroup in the relevant domain (mathematics or verbal abilities) so that
participants under threat will underperform because they may feel anxious
about confirming the stereotype or suffer from evaluation apprehension
(Steele & Aronson, 1995; Exps. 2 & 3). However, findings based on this
paradigm are open to an alternative explanation. Minorities may simply
suffer from greater evaluation apprehension or anxiety whenever a test is
diagnostic, and in principle this could occur regardless of whether the test
domain is relevant (e.g., verbal for Blacks, as in Steele & Aronson, 1995) or
not (e.g., spatial skills for Blacks) to the minority stereotype.
In a second research paradigm, participants were explicitly told that
previous research has shown differences in performance between the
minority and the majority group. In one study conducted by Spenc er et
al. (1999; Exp. 2) a group of men and women were given a difficult test of
their mathematics abilities. In the ‘‘no gender difference’’ condition
participants were explicitly told that previous research has shown that
males and females obtain similar results, whereas in the ‘‘gender difference’’
condition participants were told that differences have been found (implicitly
suggesting that women were doing more poorly than males). The results for
women showed a performance deficit in the ‘‘gender difference’’ condition
compared to the ‘‘no gender difference’’ condition, whereas males performed
equally well in the two experimental conditions. A subsequent study by
Spencer et al. (1999; Exp 3) showed that women assigned to a condition in
which no mention of the gender difference was made performed worse than
females in a ‘‘no gender difference’’ condition whereas, again, males’
performance did not vary across conditions. To put things differently,
women taking a difficult maths test showed a performance deficit compared
to men even when the maths stereotype was never mentioned whereas their
performance was eq ual to that of men when they were reassured that the
prevalent stereotype is incorrect. This study is important because it suggests
an explicit threat is not necessary for stereotype threat effects to occur and
that the maths test per se may implicitly activate the stereotype concerns
hypothesised by the model. However, these studies do not clarify whether
the performance deficit is obtained only when the task is relevant to the
stereotype, which is a crucial feature of the stereotype threat model. In
principle, women may underperform in any situation in which they are
compared to men or in any difficult testing situation unless they are
reassured about the lack of gender differences; if this were the case, being
tested in grammar, for example, may in principle repres ent a threat for
women, thus leading to a decrease in performance. This is quite different
from the stereotype threat model predicting performance deficits specifically
in those domains in which minorities are negatively stereotyped (for
STEREOTYPE THREAT 3
example, for women in maths but not in verbal skills). At this point the
literature had not demonstrated that stereotype relevance is crucial for
obtaining a stereotype threat effect.
In a third research paradig m, before performing a verbal test,
participants under stereotype threat were asked to report their race, a
request omitted in the control condition (Steele & Aronson, 1995, Exp. 4).
Similar to the non- diagnostic condition used in the other experiments, in
both conditions the test was described as a simple exercise. Again, a decrease
in performance was found under stereotype threat. This study has important
implications because it suggests that simply making salient the minority’s
social category (in this case race) will lead to stereotype threat effects. In
other words, a very subtle threat can lead to results similar to those obtained
under explicit threat. However, again, one cannot exclude that the activation
of race could lead to a generalised deficit in the performance of minority
members, so that a de crease in performance could be observed even on tasks
that are irrelevant to the minority stereotype.
Taken together, all three paradigms share one important limitation: They
do not provide convincing support for the idea that the performance deficit
observed in minority members is due to stereotype threat rather than to a
more general performance deficit, possibly due to a process of evaluation
apprehension in any testing situation and/or whenever the group stereotype
is activated. In other words, one cannot exclude the possibility that
describing a test as diagnostic, or presenting a difficult test to low-status
participants, could lead to a generalised deficit in performance, i.e., depress
performance even on tasks irrelevant to the group stereotype. Yet, the
stereotype relevance is a key feature of Steele and Arons on’s model,
according to which performance decrements are driven by ster eotype
activation and hence are limited to stereotype-relevant domains. To exclude
the possibility that people belonging to minorities are simply more
vulnerable in any testing situation, one pos sibility is to keep the
diagnosticitiy information constant and vary the na ture of the test, i.e., its
relevance to the stereotype.
An important step in this direction was made by Inzlicht and Ben-Zeev
(2000; Exp. 1). In their study, a group of women performed both a verbal
and a maths test either in a gender-balanced context or in a situation in
which females were numerically inferior to males. The ratio nale behind this
study is the following: Because unbalanced gender compositions tend to
make gender salient, the activation of the gender stereotype will be more
likely in the unbalanced than in the balanced gender composition.
Consistent with these predictions, results showed a dec rease in maths
performance for females in the minority compared to the sex-balanced
condition (for conceptually similar results see Sekaquapteka & Thompson,
2003, who found that women who had a solo status in an otherwise all-male
2
4 MAASS AND CADINU
group performed more poorly on a maths test than those workin g in an all-
female group). Most important, the gender composition of the group did
not affect their performance on the verbal test, suggesting that women’s
performance is impaired only when the test domain is relevant to the
negative stereotype about the ingroup, i.e., the maths domain. Although this
study is very informative because it suggests the importance of the
stereotype domain in interaction with gender salience, the use of distinctly
different test s leaves the stereotype-relevance question open to alternative
interpretations, such as differential test difficulty. A similar paradigm was
used by Leyens, De
´
sert, Croizet, and Darcis (2000) who had men and
women perform an affective and a valence task of similar levels of difficulty
(as well as an easier lexical decision task). Before the tasks, in the stereotype
threat condition, participants were told that men are not as good as women
in dealing with affect and processing affective information, whereas in the
control con dition the tasks were simply described as aimed at understanding
cognitive factors involved in the processing of verbal information. Males
made more errors in the threatening condition but only on the affective task,
i.e., the task that was relevant to the previous threat manipulation. This
study shows very clearly that an explicit threat about men being inferior to
women in the domain of affect was effective in decreasing men’s affective
performance.
Even more convincing is a paradigm used by Stone, Lynch, Sjomeling,
and Darley (1999; Exp. 1), who varied the test label while keeping the test
constant. Stone and collaborators told Black and White participants that
they would perform a sports test based on a game of golf. Black
participants’ performance was worse when the task was framed as
diagnostic of ‘‘sports strategic intelligence’’ (stereotype threat condition
for Blacks) compared to a condition in which the test was framed as
measuring psychological factors correlated with ‘‘general sports perfor-
mance’’ (control condition). At the same time, White participants performed
worse when the golf task was framed as diagnostic of ‘‘natural athletic
ability’’ (stereotype threat condition for Whites) than in the control
condition. In other words, although the domain of the task was kept
constant, i.e., golf, the task was presented as either testing athletic talent or
strategic intelligence (or simply as correlated with general sports perfor-
mance, in the control condition). What makes this research paradigm very
interesting is the constanc y of the test across conditions and the subtlety of
the experimental manipulation in addressing pre-existing stereotypes, the
athletic talent condition being relevant to a negative stereotype regarding
Whites (and a positive stereotype regarding Blacks) and the opposite being
true for the strategic intelligence condition.
A similar logic was used in one of our experiments (Cadinu, Maass, &
Lombardo, 2002) in which we manipulated the stereotype relevance of the
STEREOTYPE THREAT 5
test by varyi ng the test domain. We focussed on a very broad performance
domain that is closely linked to people’s self-definition as well as to their
success in life: intelligence. A group of men and women were given the same
logic test that was presented as measuring either logical intelligence or social
intelligence. Notice that widely held stereotypes associate women with
relatively poor logical intelligence and men with relatively poor social
intelligence. Results in the two conditions that are of interest here showed
that women performed worse when the test allegedly measured logical rather
than social intelligence, whereas the opposite was found for men. Thus, both
in the study by Stone et al. and in the study by Cadinu et al. Steele and
Aronson’s model was su pported within an experimental paradigm that (a)
kept the test constant but implicitly varied the stereotype relevance of the
test for each of the two groups, and (b) did not explicitly mention
stereotypes regarding one group’s inferiority. Because the decrease in
performance was only observed in the domain in which a given group is
negatively stereotyped, these studies provide strong evidence in support of
Steele and Aronson’s hypothesis that performance deficits occur specifically
in stereotype-relevant domains .
An alternative paradigm was developed by Shih, Pittinsky, and Ambady
(1999; see also Ambady, Shih, Kim, & Pittinsky, 2001; Cheryan &
Bodenhausen, 1999) who manipulated the salience of participants’ so cial
identity (gender vs ethnic) and assessed its effects on performance. In line
with the stereotype threat model, Shih et al. found that Asian-American
women performed more poorly in a maths test when their gender identity
rather than their Asian identity was made salient. Because mathematics is
related to negative stereotypes about women and positive stereotypes about
Asians, this study provides strong evidence that the stereotype consistency
between the task and the relevant social identity is the key factor triggering
the decrease in performance.
This brief overview suggests that stereotype threat effects emerge
consistently across very different experimental paradigms. In particular,
depending on the experimental paradigm, stereotype threat is activated in
very different ways, for example by varying test diagnosticity, by activating a
socially shared stereotype e ither explicitly or implicitly, by activating the
minority’s category member ship prior to test taking, or by varying the test
label and, hence, its stereotype relevance. Also, stereotype-induced
performance deficits have been observed for a variety of minority groups,
including ethnic minorities (Stone et al., 1999), women (e.g., Spencer et al.,
1998), people with low socio-economic status (Croizet & Claire, 1998),
elderly (Maass, Cadinu, & Verga, 2002), and Southern Italians (Mezzapesa,
1999 2000). Apparently, then, stereotype threat is a very robust phenom-
enon that occurs across different research paradigms, different minority
groups, and different stereotypes.
3
4
5
6 MAASS AND CADINU
Since published studies are not always a reliable index of robustness, we
also looked at dissertation abstracts (less susceptible to ‘‘file draw er losses’’
than publ ished work). Considering the dissertation s on stereotype Threat
reported during the last 3 years, the ‘‘success’ rate is surprisingly high. Out
of 18 dissertations that manipulated stereotype threat and measur ed
performance, 11 reported reliable performance decrements as a function
of stereotype activation and 2 additional studies provided partial support.
Only 3 out of 18 dissertations failed to find stereotype threat effects. Also,
stereotype threat effects seem to occur to similar degrees across different
minority groups including women, racial or ethnic minorities (Blacks,
Mexican American, Jews, etc.), gays, and the elderly. Together, these studies
suggest that stereotype threat effects may be quite pervasive.
IS STEREOTYPE THREAT SPECIFIC TO LOW-
STATUS MINORITIES?
Several studies have addressed the issue of whether stereotype threat deficits
are limited to low-status minority members or can also be observed in the
performance of high-status majority members. Resul ts show that, when
negative stereotypes about the ingroup are made salient, performance
deficits can also be observed on the part of people belonging to dominant or
high-status groups. In a study conducted by Aronson, Lustina, Good,
Keough, Steele, and Brown (1999) a group of Americans of European
descent was told that Asians tend to perform better than Whites in tasks
involving mathematics abilities. Participants under threat showed a decrease
in performance compared to a control condition in which Asians were not
mentioned. Similar results were found by Leyens et al. (2000) in a study
discussed earlier. The results showed that men underperformed in affective
tasks when their inferiority to women had explicitly been stressed. Both of
these studies investigated performance deficits of majority members but did
not include minority members. Thus, one question that remains open is
whether the stereotype threat deficits suffered by low-status and high-status
group members are similar in size. In the research conducted by Cadinu et
al. (2002, Exp. 1), stereotype threat effects suffered by minority and majority
members were assessed within the same experimental design and results
showed that the effects for male participants (under social intelligence
threat) were as strong as the results found for female participants (under
logical intelligence threat).
Overall, these studies suggest that a history of stigmatisation is not
necessary for stereotype threat effects to occur. Moreover, the study by
Cadinu et al. (2002) suggests that the stereotype threat phenomenon may be
just as strong for dominant social groups as for stigmatised or low-status
groups. Yet the processes underlying performance deficits by high- vs low-
STEREOTYPE THREAT 7
status groups may not be the same. In a recent study by Cadinu, Maass,
Frigerio, Impagliazzo, & Latinotti (2003a), a group of Black American
soldiers living in Italy were exposed to a manipulation that rendered either
their race (low-status group) or their na tionality (dominant group) salient.
They were told that their in-group generally fared either very well or rather
poorly on verbal abilities compared to the outgroup (compared to Whites
when the Black identity was salient or compared to Italians when the
American identity was salient).
1
Subsequently, they were asked to report
their level of expectation regarding their performance on a verbal
intelligence test and later performed the test itself. Note that the test was
the same in all experimental conditions. Consistent with predictions,
participants both in the Black (low-status) and in the American
(dominant-status) condition had lower expectations and under-performed
after having received negative information about the in-group. However, the
level of expectancy was found to mediate the decrease in performance for
participants in the Black but not in the American condition. These results
suggest that, although comparable performance deficits were found for
members in the low-status (Black) and in the dominant-status (American)
group, the underlying processes may be different. One possibility is that, in
the case of dominant or high-status group members, the stereotype threat
effects are simply the result of a temporary situational pressure whereas for
low-status minority members a history of stigmatisation leads these
individuals to lower their level of expectation which in turn causes the
performance decrement. Although no definite conclusions can be drawn on
the basis of a single study, the findings by Cadinu et al. (2003) suggest that
caution is needed when extending a general stereotype threat model to high-
status group memberships because the processes underlying comparable
performance deficits may be of quite different nature.
MODERATOR VARIABLES: WHO IS MOST
VULNERABLE TO STEREOTYPE THREAT?
Although stereotype threat appears to be a robust and widespread
phenomenon, people are not equally susceptible to its debilitating effects.
To date, a number of individual difference variables have been identified
1
In a pilot test conducted in the US, 22 non-Black participants were asked to indicate which
group (Blacks vs Whites and Americans vs Italians) were generally believed to have better
verbal abilities. Whites were associated with better verbal abilities than Blacks (3.55 on a 5-point
scale, which different significantly from the neutral midpoint of 3, one-sample t-test = 3.46,
p 5 .01) and Italians with better verbal abilities than Americans (3.27, different from neutral
midpoint of 3, one-sample t-test = 2.02, p 5 .06). Thus, our Black American soldiers apparently
belong to a group whose verbal abilities are negatively evaluated both on the basis of race and,
to a slightly lesser degree, nationality.
8 MAASS AND CADINU
that render minority members more or less vulnerable to stereotype threat,
including domain identification, group identification, and internal vs
external control beliefs.
One individual difference variable that is known to moderate stereotype
threat effects is the importance that individuals attribute to the relevant
performance domain. Starting from Aronson et al.’s research (1999; Exp. 2),
there is fairly consistent evidence that a strong identification with the
relevant domain constitutes a risk factor that makes people particularly
vulnerable to stereotype threat (e.g., Cadinu et al., 2003a, Exp. 1; Spencer et
al., 1999; Stone et al., 1999, Exp. 1). Those who consider the performance
domain important (and who generally tend to be good at it) are the ones
that are most strongly affected by stereotype activation, whereas their less
motivated and less competent peers seem almost immune to stereotype
threat. Indeed, domain identification is considered a necessary conditio n for
stereotype threat to occur and many authors now exclude a priori from their
experiments those participants who are not highly identified with the
performance domain (cf, Marx & Roman, 2002). Although the reasons for
the differential susceptibility remain unclear, one plausible explanation is
that people with low domain identification may have withdrawn from the
domain sometime during their life, possibly as a consequence of repeated
experiences with stereotype threat . Their dissociation from the domain may
then protect them from any further stereotype threat experience exactly
because the performance domain is no longer relevant to their self-concepts.
We will come back to this possibi lity when discussing the mechanisms
driving stereotype threat.
A second individual difference variable that seems to moderate stereotype
threat is the degree to which people identify with the social group to which
they belong. Drawing on social identity theory, Schmader (2002) argued
that stereotype threat should mainly affect those who are highly identified
with the group that is the target of the stereotype. Indeed, in line with this
prediction, he found that only women strongly identified with their gender
group suffered a performance decline when taking a mathematics test aimed
at investigating gender differences. When the test was described as
measuring individual abilities (rather than gender differences), women
performed just as well as men, regardless of their gender identification.
Thus, group identification appears to be an important moderator of
stereotype threat, possibly because highly identified individuals feel more
apprehensive about confirming a negative stereotype regarding their ingroup
(although some other experiments have failed to confirm this prediction, see
Cadinu et al., 2003).
A third variable that appears to moderate stereotype threat is locus of
control. Cadinu et al. (2003a, 2003b) have argued that individuals with
internal locus of control would be more vulnera ble to the effects of
6
STEREOTYPE THREAT 9
stereotype threat than those characterised by an external locus of control.
Individuals with internal locus of control tend to believe that their successes
and failures depend on themselves. The fact that they feel personally
responsible for their actions, in conjunction with the pressure deriving from
the stereotype threat condition, will put these individuals under intense
pressure, which in turn wi ll impair performance (see Svenson & Maule,
1993, for evidence that excessive stress is related to low performance). In line
with this reasoning, Cadinu et al. (2003a) found that the performance
expectancies of internal control individuals were greatly reduced when
exposed to stereotype threat, while external control individuals were
unaffected by stereotype activation. However, no corresponding effects
were found for actual performance in this study. In contrast, Cadinu et al.
(2003b) found that internal locus individuals underperformed under
stereotype threat compared to external locus individuals. This finding is
interesting in view of the fact that students with internal control beliefs have
generally been found to perform better than those with external control
beliefs (Pintrich, Marx, & Boyle, 1993). Despite this general advantage,
individuals with internal locus seem to be at greater risk when expo sed to
stereotype threat, as evidenced for performance expectancy in Cadinu et al.
(2003a) and for actual performance in Cadinu et al. (2003b).
Together, these studies suggest that some individuals should be
considered at high risk for stereotype-related performance deficits, including
those who consider the performance domain particularly important, those
who identify strongly with their stigmatised ingroup, and those who have
internal control beliefs. The role of other potential moderator variables such
as high stigma consciousness (Pinel, 2002) remain to be explored.
THE DEVELOPMENT OF STEREOTYPE THREAT
When do stereotype threat effects arise and how early will minority children
suffer stereotype-related performance deficits? These questions are extremely
important if we want to understand (and possibly prevent) the development
of stereotype-driven disadva ntages. For stereotype threat effects to occur,
children need to have acquired three types of knowledge. First, they need to
have developed a concept of social categories (category awareness), second,
they need to be able to confidently identify themselves as members of a given
category (self-categorisation), and third, they need to be aware of the fact
that specific categories are positively or negatively associated with specific
domains (stereotype knowledge). When will children meet these criteria? As
far as the first prerequisite, category awareness, is concerned, there is
evidence that children learn to perceptually distinguish categories such as
males vs females by the end of their first year (Fagot & Leinbach, 1993). By
age 3, the majority of children are not only able to label social categories
7
6
6
10 MAASS AND CADINU
correctly but they have also developed a basic awareness of status
differences between social groups and start to show ethnic and gender
preferences. Thus, although children of that age may not have full
understanding of the defining features and the constancy of social groups,
they have acquired a basic understanding of social categories (Fagot &
Leinbach, 1993).
Turning to the second prerequisite, self-categorisation, 2-year olds still
tend to experience uncertainty in defining their own sexual or racial group
membership, but by abo ut 3 years of age, children are generally able to
confidently identify their own category membership (e.g., Fagot &
Leinbach, 1993; Thompson, 1975). Finally, considering stereotype knowl-
edge, there is evidence that, even before 3 years of age, children have some
idea about how members of different categories, such as boys and girls, are
expected to behave. By the age of 4, they are able to correctly understand
specific stereotypic expectancies (Killen, Pisacane, Lee-Kim, Ardila-Rey,
2001; Signorella, Bigler, & Liben, 1993). It is interesting to note that children
adopt stereotype-consistent (e.g., gender-appropriate) behaviours long
before having mastered group (e.g., gender) constancy (see Fagot &
Leinbach, 1993, and Martin, 1993, for overviews). Not surprisingly, such
stereotype knowledge is limited to age-relevant domains such as toy/game
preferences, appearance, athletic abilities and preferences, future profes-
sions, etc; it may not include domains such as mathematics or astrophysics
that are beyon d the experience of small children. Despite this obvious limit,
there is no reason to believe that, in principle, children are una ble to
experience stereotype threat. By the time they enter schoo l, children have
long acquired all of the cognitive prerequisites of stereotype threat including
category knowledge, self-cat egorisation, and stereotype knowledge.
Although it is clear that children acquire these prerequisites very early on,
surprisingly little is known about the ways in which this knowledge may
affect their performance expectations and their actual performance in
stereotype-relevant domains. There is evidence that boys’ and girls’
achievement beliefs differ in ster eotypic ways as early as first grade, and
that these beliefs reliably affect subsequent performance. In particular, some
studies found that girls alrea dy show lower performance expectations in
mathematics in their first schoo l year (Eccles, Wigfield, Harold, &
Blumenfeld, 1993; Wigfield et al., 1997). Also, both parents’ (Tiedemann,
2000) and teachers’ gender stereotypes (Keller, 2001) seem to influence the
performance expectancies of elementary school children.
More telling about stereotype threat effects in children is a recent series of
studies by Ambady et al. (2001) suggesting that stereotypes may directly
interfere with performance in relevant domains. In the first study, Asian girls
attending elementary or middle school performed worse on a mathematics
test when their gender category had been made salient prior to the test than
STEREOTYPE THREAT 11
in a no-salience control condition. This was most pronounced for 5-year-
olds (K-grade) and for middle school children, whereas upper elementary
school children showed a different pattern. Luckily, this same study also
suggests that stereotypes may, at times, faci litate performance. An intriguing
feature of this study was that in another condition it was the Asian (rather
than gender) category that had been activated. In this condition, these girls
reformed reliably better than in the control condition. In other words,
already by 5 years of age, stereotypes may exert either a debilitating or
facilitating effect on performance, depending on the positive (Asian
mathematics) or negative (females mathematics) association between
performance domain and social category.
Another, as yet unpublished, study by Good (2001) suggests that gender
stereotypes may gain importance with increasing age. Comparing fourth,
fifth, and sixth graders, Good found that older girls were particularly
concerned about inadvertently confirming the stereotype that associates girls
with poor maths performance. Indeed, these older (sixth grade) girls
performed worse than boys under stereotype threat, but they actually
outperformed boys in the absence of such threat. Taken together, the few
available studies on school-aged children suggest that girls may suffer from
stereotype threat effects as early as 5 or 6 years of age and that the
magnitude of this phenomenon may reach a peak in middle school children
(between grade 6 and grade 8). Even very subtle form s of category activation
(such as colouring a picture of a girl) seem sufficient to reliably affect
performance in stereotype-relevant domains. Unfortunately, nothing is
known so far about stereotype-induced performance deficits in children
belonging to other stigmatised groups such as Blacks, Hispanics, or Gypsies
nor is there any evidence that stereotype threat -induced impediment (or
facilitation) is linked to other aspects of cognitive development.
WHAT PROCESSES ARE DRIVING STEREOTYPE
THREAT?
While there is little doubt about the pervasiveness and early onset of the
phenomenon, the primary challenge currently faced by research ers is the
identification of the mechanisms underlying stereotype threat. Differ ent (and
often complementary) hypotheses have been formulated, but to this point no
single mechanism has emerged as clearly driving stereotype threat.
Anxiety
The first explanation, originally proposed by Steele and Aronson (1995),
suggests that minority members are afraid of confirming the stereotype that
associates their own group with poor performance in a given domain. In
12
MAASS AND CADINU
addition to ‘‘normal’’ or baseline anxiety associated wi th taking difficult
tests, minority members may experience further tension due to their
preoccupation with inadvertently confirming a negative stereotype. Since
high levels of anxiety and arousal have generally been found to be
detrimental to task performance on difficult tasks (e.g., Hill & Wigfield,
1984), it is not surprising that stereotype-induced increases in anxiety will
interfere with performance. Although the performance-debilitating effects of
anxiety are well establ ished in the literature (see Dembo & Eaton, 1997, for
an overview), in the earlier work on stereotype threat no effects of stereotype
threat on anxiety were found. For example, Steele and Aronson (1995, Exp.
2) measured participants’ levels of anxiety immediately after the test and
found no effects of stereotype threat on anxiety. Similarly, Aronson et al.
(1999) found no effects of stereotype threat on a measur e of anxiety taken at
the end of the test. Stone et al. (1999) measured anxiety before and after the
test and found a general increase in anxiety but this increase was no stronger
in the stereotype threat than in the control condition. However, a more
promising finding was reported by Spencer et al. (1998, exp. 3), who found
that stereotype threat had a margi nal effect on anxiety which in turn had a
negative effect on performance.
Two recent studies have also provided suggestive evidence that anxiety
may play a role in stereotype threat . Osborne (2001) found that race
differences in test performance were mediated by anxiety. However, no
strong conclusions can be drawn from this study because the mediator
variable (anxiety) was measured after the outcome variable (performance).
More telling is a study by Blascovich, Spen cer, Quinn, and Steele (2001) who
found that African Americans exhibited larger increases in mean arterial
blood pressure under stereotype threat than European Americans , a finding
that is clearly in line with the anxiety hypothesis. Thus, although empirical
proof is limited at this point, there is preliminary evidence that anxiety may
play a causal role in stereotype threat.
Intrusive thoughts
A second way in which the fear of confirming negative stereotypes may
interfere with performance is via intrusive thoughts. When put under
stereotype threat, minority members are subject to dual cognitive demands:
on one side, their cognitive resources are occupied by a challenging task, on
the other side, they are occupied by thoughts related to the activated
stereotype. Thoughts about inadvertently confirming the stereotype are
likely to distract the minority member from what should be his or her
primary task. So far, there is some evidence that the stereotype is cognitively
activated under stereotype threat (see Steele & Aronson, 1995, Exp. 3); what
remains to be demonstrated is that stereotype activation reduces cognitive
4
STEREOTYPE THREAT 13
resources available for the primary task, as hypothesised in Steele’s model.
Although the divided attention hypothesis remains, in our opinion, a strong
candidate for explaining stereotype threat, it is still awaiting empirical
confirmation.
Shift towards caution
A third possibility is that the activation of the stereotype will increase
people’s evaluation apprehension in the stereotype-relevant domain which in
turn will lead them to be more cautious when providing responses. In other
words, the concern about how they will be evaluated, together with a desire
to avoid failure, will motivate minority members to exert excessive caution
when performing the task. This explanation differs from the previous ones in
an important way: Rather than focusing on reduced problem-solving
capacities under stereotype threat, this approach hypothesizes a criterion
shift when guessing.
Although there is only weak evidence for the mediating role of evaluation
apprehension in stereotype threat (e.g ., Aronson et al., 1999; Spencer et al.,
1999), there is some support for the idea that participants under stereotype
threat are reluctant to take risks when responding, resulting in a high
number of blank responses (e.g., Aronson et al., 1999, Exp. 2; Croizet &
Claire, 1998: Steele & Aronson, 1995). For instance, an unpublished thesis
by Mezzapesa (1999 2000) conducted in collaboration with our research
team supports the hypothesised shift towards caution. In this study,
Southern Italian high-school students took a difficult logical-mathematical
test under three different conditions. In the c ontrol condition, no mention
was made of their category membership. In the explicit threat condition,
they were told that Southern Italian students generally performed worse on
this kind of test than Northern Italians. Most importantly, in the implicit
threat condition they were required to identify their own category
membership (‘‘Southern’’ vs ‘‘Northern’’) prior to the test, but no explicit
mention was made of performance differences. This manipulation is believed
to activate, in a subtle way, the widely held stereotype that associates
Southern Italians with lower status and lower performance compared to
Northern Italians. The results showed that the number of correct responses
was reliably lower under stereotype threat, especially when threat was of the
implicit kind. These same participants also provided fewer responses and left
more problems blank. However, when we looked only at those items for
which participants attempted to give a response, then the success rate was at
least as good as that in the control group. In other words, threatened
students did not make any more mistakes, but they were much more
cautious in providing responses and indeed failed to provide responses on a
great number of items. Presumably, participants under stereotype threat
14
MAASS AND CADINU
were more reluctant to provide ‘‘educated guesses’’ whenever they felt
uncertain about their responses. Obviously, these data are open to
alternative explanations such as effort withdrawal and hence are in need
of additional testing. Despite this limit, these findings are in line with the
idea that the performance decrement of threatened individuals may be
attributable to a more cautious decision criterion rather than to poorer task
performance.
Interestingly, Leyens et al. (2000) found a very different pattern for
dominant group members exposed to stereotype threat. Threatened men
showed impaired performance under threat, attributable to c ommission
(rather than omission) errors. When asked to identify affective words, they
shifted their threshold criterion, tryin g as hard as possible not to miss any
such words; this resulted in a great number of false positive responses. One
possibility is that low-status and high-status group members use different
coping strategies under stereotype threat, the former becoming more
cautious, the latter taking greater risks. However, because caution and risk
taking have been operationalised differently in the different studies, no
conclusion can be drawn at this point.
On the methodological side, the results of the above studies suggest that
researchers should construct performance measures so that erroneous
responses can be clearly distinguished from non-responses, as they may
reflect distinct psychological processes.
Performance expectancy
A fourth potential mechanism of the stereotype threat effect is the
individual’s level of expectancy regarding his or her performance. According
to this perspective, stereotype threat may decrease performance expectancies
that in turn may impair performance. Such a self-fulfilling cycle assumes (a)
that stereotype threat reduces minority members’ level of expectancy
regarding their task performance and (b) that reduced expectancies will
produce a performance decreme nt. Dis tinct from other explanations of
stereotype threat, this explanation focuses on the initial stage in which
people evaluate the subjective likelihood of success depending on their
personal resources. Because personal resources are often anchored to group-
level expectations, ingroup-threatening information (e.g., ‘‘women are less
competent in maths’’) may reduce personal expectancies about doing well.
In line with this idea, Stangor, Carr and Kiang (1998) found that
participants had consistently lower performance expectancies under stereo-
type threat even if they had received positive feedback on a previous test.
However, Stangor et al.’s research (1998) does not provide any information
about the link between expectancy and performance as no actual
performance data were collected. To date, there are only a few studies that
STEREOTYPE THREAT 15
have investigated the potentiall y mediating role of performance expectancy
in stereotype threat, but findings are not entirely consistent.
The link between expectancy and performance was addressed in two
unpublished studies (reported in Steele & Aronson, 1995) in which
performance expectations were experi mentally manipulated after stereotype
threat, but no effects on performance were found. A problem with
experimental expectancy manipulations is that they may not only conflict
with the stereotype threat manipulation, but they may also be too weak to
override the participant’s personally held beliefs. Rather than manipulating
expectancies experimentally, we have recently focused on the participants’
personal performance expectancies and have tried to untangle their relation
with performance (Cadinu et al., 2003a). In our first study, female university
students were assigned to one of three experimental conditions. They were
either told that women generally perform worse on logical-mathematical
tests than males (stereotype threat conditio n), or that they perform a s well as
males (control condition), or that women generally outperform males
(counter-stereotypical information).
2
Later, they were given a difficult maths
test and asked to estimate their performance prior to taking the test. Not
surprisingly, participants who considered mathematics unimportant were
not affected by the manipulation. In contrast, participants who considered
mathematical-logical abilities an important domain were highly sensitive to
the manipulation both with regard to perfor mance expecta ncies an d actual
performance. They were most optimistic about their future performance
after receiving counter-stereotypical information and least optimistic under
stereotype threat. Consistent with these expectations, they did indeed
perform best after counter-stereotypical, and hence ingroup-favouring,
information but showed a sharp decline in performance under stereotype
threat. Most important, mediational analyses suggest that actual perfor-
mance was partially mediated by expectancies.
A second experiment, reported in an earlier section of this chapter,
involved Black Americans living in Italy. These participants were either
made aware of their Black identity, a category with a long history of
stigmatisation, or of their American identity, a high-status category. In both
cases, participants were given either favourable or unfavourable information
about how well Blacks (compared to Whites) or Americans (compared to
2
To increase t he credibility of the manipulation, participants were told that research
investigating performance differences in different areas of cognition generally finds no
differences between men and women. However, there are some areas in which differences do
emerge. One of them concerns logical-mathematical abilities where women obtain higher (or
lower, depending on condition) scores than men. To underline the veridicality of this statement,
a fictional research study was referred to and a graph was shown supposedly summarising ‘‘the
results obtained by men and women based on 72 studies, 36 of which employed randomized samples
of men and women (Adapted from Taylor, Sheatsley, & Greeley, 1998)’’.
16 MAASS AND CADINU
Italians) performed on verbal tests. Regardless of whether their Black or
American identity had been activated, participant s had lower expectations
(see Figure 1) and under-performed after receiving negative informat ion
about the in-group (see Figure 2). However, the level of expectancy was
Figure 1. Expectancy as a function of Category Membership (Black vs American) and Type of
Information (positive vs negative) (Cadinu et al. 2002a, Exp. 2).
Figure 2. Performance as a function of Category Membership (Black vs American) and Type of
Information (positive vs negative) (Cadionu et al. 2002a, Exp 2).
STEREOTYPE THREAT
17
found to mediate the decrease in performance for participants in the Black
but not in the American condition. Together, these two studies suggest that
pessimistic performance expectancies may be one mechanism through which
stereotypes induce performance deficits. At the same time, both studies
suggest that expectancies are not the only process mediating stereotype-
related performance decline, considering that there was evidence for partial
but not for full mediation. As discussed earlier, our second study has
another implication, namely that apparently identical performance deficits
in low- (women, Blacks) and high-status groups (men, Whites) may in
reality be driven by distinct processes. When there is a history of
stigmatisation, ingroup-threatening information (or the mere activation of
the category) may be sufficient to induce reduced expectancies which in turn
lead to reduced performance. In high status groups, expectancies seem to
play no mediating role in threat-induced performance decrements. Thus,
although comparable performance deficits are found for low- and high-
status group members, the underlyin g processes may be quite different.
Disengagement and effort withdrawal
Whereas the four processes discussed so far deal mainly with momentary
stereotype threat, a fifth and last mechanism, disengagement and effort
withdrawal, focuses on long-term effects. According to the stereotype threat
model, minorities may, in the long run, ‘‘disidentify’’ from the relevant
performance domain. Distancing oneself from a domain fulfills a self-
protective function (‘‘maths is irrelevant to my self-concept’’) but it also
implies effort withdrawal and, hence, poor performance. For example,
women asked to perform a typically masculine task, such as playing chess,
may have learned to consider this domain irrelevant, possibly due to
repeated stereotype threat experiences. As a consequence, they may put very
little effort into solving a difficult chess problem. Ironically, these women (or
low-status group members in general) are not particularly vulnerable to the
momentary activation of the stereotype, exactly because they have already
disengaged from the domain in prior stages of their learning history. As
mentioned earli er, indirect support for this hypothesis comes from different
studies showing that stereotype threat effects were only present for
participants highly identified with the testing domain (Cadinu et al.,
2003a, Exp. 1; Spencer et al., 1999). In contrast, participants with high
involvement in the domain (for example women who consider chess
important and indica tive of their intellectual capacities) will suffer
momentary stereotype threat effects. It is impor tant to point out that the
stereotype threat-induced performance deficit of these women or minority
members cannot be attributed to effort withdrawal. To our knowledge, there
is no evidence to date that these participants put less effort into the task or
18
MAASS AND CADINU
that they are less persistent during task performance. Apparently, these
women underperform for reasons other than effort withdrawal. Never -
theless, it would be important for future research to measure participants’
level of effort and persistence on the task.
In general, demotivation and reduced effort seem to be typical of those
low-status group members who have withdrawn from the domain but it does
not seem to account for performance deficits of highly involved minorities
under momentary stereotype threat. Although the long-term effects of
stereotype threat are very interesting, we will not consider them any further
as they are beyond the scope of this chapter.
Taken together, there is no definite answer as to which mechanisms are
driving stereotype threat. Different authors have hypothesised and tested
different potential mediators, but to date none of these has received
unequivocal support. To complicate things further, the results obtained so
far suggest that not all stereotype threat effects may be attributable to a
single process. Thus, we may be faced with a multifaceted phenomenon that
is the joint function of different psychological mechanisms.
RELATION TO OTHER THEORIES
More than other sciences psychology has at times been accused of
childhood amnesia. Principles and phenomena that had been known in
the early years of psychology are no longer accessible to the collective
memory of the discipline and are therefore ‘‘rediscovered’’ under new
headings and enthusiastically greeted by the forgetful scientific
community. At first glance, the stereotype threat model seems to be
such a case. Self-fulfilling prophecies have been known to sociologists
as well as social and educational psychologists since the 1940s (Merton,
1948). In particular, resear ch conducted on the Pygmalion effect during
the 1960s (Rosenthal, 1994; Rosenthal & Jacobson, 1968) has
demonstrated the powerful effect of teacher expectations on the
(objectively measured) performance of their students. More recently,
Eccles and collaborators have extended and integ rated these findings
within expectancy-value theory of achievement motivation (cf, Eccles,
Wigfield, & Schiefele, 1998; Wigfield & Eccles, 2000). Considering this
long tradition, one may suspect that stereotype threat is nothing more
than a specific case of behavioural stereotype confirmation. We will
briefly examin e this possibility as it is important to understand to
which degree the stereotype threat mod el contributes beyond previous
ones. In particular, there are two previous models, self-fulfilling
prophecy and expectancy-value model, that seem to cover the same
ground, and a further model, attributional ambiguity, that appears to
be a contrasting approach.
STEREOTYPE THREAT 19
Self-fulfilling prophecy
Looking first at self-fulfilling prophecies (Jussim, 1986; Snyder & Haugen,
1995; Snyder & Stukas, 1999), this phenomenon is generally conceptualised
as a three-step chain of events. First, person A holds expectations about a
target person B that are often (although not exclusively) based on
stereotypic beliefs about B’s category membership, including race, gender,
religion, social class, nationality, etc. Second, these expectations guide A’s
behaviour in interaction wi th person B. Third, person A’s behaviour, in
turn, influences B’s behaviour in a way that confirms the initial expectations
of person A. Importantly, the target person B is generally una ware of both
person A’s expectancies and of changes in his/her own behaviour (cf,
Vorauer & Miller, 1997). Also, research has consistently shown that it is the
low-status or low-power target person who inadvertently confirms the
expectancies held by a powerful interaction partner (cf, Copeland, 1994; see
Snyder & Stukas, 1999, for an overview). This situation is quite similar to
the stereotype threat situation where it is generally the member of a
stigmatised minority who suffers the greatest performance deficits. Thus, in
many ways, the self-fulfilling cycle described by Snyder and collaborators
seems to resemble the stereotype threat model in which stereotypic beliefs
(e.g., women are incompetent in maths) produce stereotype-confirming
behaviours (decline in performance).
However, there are two important differences between these models. First
of all, the self-fulfilling prophecy model requires the presence of an
interaction partner who exerts a guiding influence and elicits reactions from
the target person that conform to initial expectations. Hence, the behaviour
of the interacti on partner A is essential in modifying the behaviour of target
person B. Stereot ypes exert an influence on the target’s behaviour only
through the mediating role of the person holding the stereotypes. This is
quite different from the stereotype threat model, according to which
stereotypes, once activated in the mind of the target (low-status group
member), can interfere with task performance even in the absence of any
other person. Thus, in both cases, stereotypes influence the behaviour of the
minority target in a self-confirming way, but the causal link between the
stereotype and the confirmatory behaviour are quite different. In the self-
fulfilling prophecy cycle, it is the prejudiced interaction partner who
mediates this effect, whereas in the stereotype threat model the activation of
the stereotype in the mind of the target is sufficient to produce changes in the
target’s behaviour or performance.
Second, the two models differ in the importance assign ed to the target’s
awareness of stereotypic expectancies. Self-fulfilling prophecies generally
occur when the target person is unaware of the negative expectancies. When
the target person realises that the interaction partner hold s negative
20
MAASS AND CADINU
expectancies, s/he usually makes attempts to disconfirm these expectancies.
There is now considerable evidence in the self-fulfilling prophecy literature
suggesting that targets of stereotyping are likely to engage in behavioural
dis-confirmation if they are aware of the stereotypic expectancies held by
others (e.g., Hilton & Darley, 1985; Stukas & Snyder, 2002). Thus, self-
fulfilling prophecies req uire that the target be unaware of the negative
stereotypic expectancies. To the contrary, stereotype threat effects require
awareness. Indeed, stereotypes will not affect performance unless they are
activated in the test situation. Taken together, the two models both
hypothesise a self-sustaining cycle of stereotypes, but they assume different
underlying mechanisms and operate under different boundary conditions.
Expectancy value theory
The second model that seems to overlap with the stereotype threat model is
expectancy value theory. According to this model, achievement-related
choices and performance are determined by both subjective task value and
expectation of success which, in turn, depend on a great number of predictor
variables (see Wigfield & Eccles, 2000, for a recent version of the model).
Among these variables are stereotypes. The culturally shared stereotypes,
transmitted during socialisation especially by parents, affect the child’s self-
schemata, goals, and, in particular, his/her ability, beliefs and subjective
probability of success in stereotype-relevant domains. Thus, expectancy
value theory emphasises, among others, the role of expectancy in
determining the level of performance (Eccles et al., 1998; Wigfield, 1994;
Wigfield & Eccles, 1992). In this model, people’s performance can be
explained (a) by their beliefs about how well they will do on a given task and
(b) by the extent to whi ch they value the task. Both of these are, in part,
determined by stereotypes.
There is obviously considerable conceptual overlap between stereotype
threat and the expectancy value model, but again, each theory has distinct
features that make it unique. The first concerns the underlying processes.
Whereas expectation of success is a critical mediator in the expectancy value
model, stereotype threat may or may not be mediated by this same process.
As mentioned earlier, there is only limited evidence that stereotype threat is
mediated by reduced expectanc ies, and different alternative explanations
remain ope n, including the possibility that the low-status group member
makes a conscious attempt to disconfirm the stereotype. This has potentially
important theoretical implications. Whereas the expectancy value frame-
work assumes that the member of a stigmatised group has incorporated the
prejudiced world view in his/her self-schemata and values, such an
internalisation is unnecessary from the perspective of the stereotype threat
model. The stigmatised person may disagree entirely with culturally shared
STEREOTYPE THREAT 21
stereotypes, but may still suffer stereotype-induced performance deficits due
to intr usive thoughts, anxiety, or a shift towards caution. Another
fundamental difference between the two lines of research is that research
within the expectancy value model generally focuses on relatively stable,
long-term effects of stereotypes (often investigated in longitudinal designs)
whereas the stereotype threat model focuses on contextually situated
stereotype activation.
In sum, the stereotype threat model seems to make a series of distinct
predictions that make it a complementary rather than a redun dant model
with respect to previous theories.
Attributional ambiguity model
Whereas the self-fulfilling prophecy and the value- expectancy model seem to
converge in many aspects with the stereotype threat model, there is another
approach that is in apparent contrast. Readers may have noticed a surface
contradiction between the stereotype threat model and earlier work on
attributional ambiguity. With reference to Dion’s (1975) and Crocker and
Major’s (1989) earlier work, Major et al. argue (in this volume) that
stereotype awaren ess may at times have beneficial consequences for low-
status group members. In situations in which stereotypes are salient and
low-status group members experience failure, they may be protected against
self-esteem loss exactly because they can attribute failure to prejudice rather
than to a personal lack of abilities.
At first glance, it may be surprising that the same conditions, namely
stereotype activation, may, on the one hand, lead to a performance deficit
(or failure) while, on the other hand, bolstering the minority’s self-esteem.
This apparent paradox represents a challenging question for future research.
Unfortunately, we are not aware of any studies in which stereotype threat-
induced performance deficits and self-esteem protection have been shown
within a single experimental paradigm. Theor etically, it is conceivable that
the two effects coexist.
At the same time, we should also point to a number of differences
between the tw o areas of study, stereotype threat and attributional
ambiguity, that render a direct comparison difficult. Besides focusing on
different aspects of human behaviour (performance vs self-esteem), the two
areas also differ in their operational definition of stereotype awareness.
Within Major et al.’s paradigm, self-esteem protection is contingent on the
belief that the evaluator holds negative stereotypes about the target’s own
group. In other words, in Major et al.’s model, real or presumed prejudice of
the evaluator is essential to protect failing minorities from self-esteem loss.
In stereotype threat research, however, stereotype awareness is defined as a
more abstract consciousness that there are socially shared stereotypes about
22
MAASS AND CADINU
one’s own group. Stereotype-induced deficits require that the stereotype is
activated in the mind of the minority test-taker and may well occur in the
absence of any presumed evaluator bias (for example, on an electronically
graded maths test). In conclusion, stereotypes have a different conceptual
status in the two models and it remains to be seen whether they can
simultaneously function as a barrier to performance and a buffer against
self-esteem loss.
Taken together, there are multiple conceptual links between stereotype
threat and other models including self-fulfilling prophecy, expectancy value,
and attributional ambiguity. This brief comparative overview suggests that
each of these models has a unique domain of application and makes distinct
assumptions about underlying processes, although one may envisage the
future development of a more general integrative model.
APPLICATIONS
The areas of application of stereotype threat are virtually unlimited.
Whenever members of stigmatised groups engage in a task that is, according
to socially shared stereotypes, negatively associated with their social
category, there are opportunities for stereotype threat to occur if the
category and/or stereotype is made salient. To date, most studies have
focused either on Blacks or on women, although stereotype threat effects
have also been observed for other stigmatised groups such as low social class
(Croizet & Claire, 1998) or elderly people (Maass et al., 2002). With few
exceptions (Stone et al., 1999), research has generally analysed stereotype
threat effects in different academic domains (maths, verbal tests, etc.), but in
principle there is no reason to believe that stereotype threat is limited to
these cases. For example, will the widely shared stereotype of female drivers
affect the actual driving behaviour of women? What will happen to the
concentration of the lonely fema le chess player faced with 40 male players in
a room where an important tournament is taking place? Although females
are quite successful in chess as youngsters, how come they are practically
absent as chess masters? Does the stereotype of chess as a typically male
activity play a role in this? As these few examples illustrate, potential (as yet
unexplored) areas of application are practically unlimited.
In our opinion, a particularly interesting and potentially controversial
application is affirmative action. There is little doubt that affirmative action
is a useful tool to enhance the number of women or minority members in
industry, public services, and educational institutions (for an overview see
Valian, 1998). At the same time, affirmative action has various negative
consequences for those hired under affirmative action rules. In particular,
these minority members tend to be less satisfied with their job (Chacko,
1982), they tend to be judged less competent by their colleagues (Heilman,
5
STEREOTYPE THREAT 23
Block, & Lucas, 1992), and they have less trust in their own abilities
(Heilman, Lucas, & Kaplow, 1990). One possible explanation of this latter
finding is that affirmative action rules activate existing gender and/or race
stereotypes. Imagine a woman taking an entrance exami nation for a
prestigious university or to participate in a job selection under affirmative
action rules . Affirmative action rules indirectly activate existing gender
stereotypes, as they offer legal and institutional protection to members of
disadvantaged groups that are subject to discrimination and presumed by
some members of society to be less capable. Applying the stereotype threat
model to this situation, we would expect women to do less well on
admission/hiring tests if they are aware of the affirmative action laws.
First evidence for this possibility come from a series of studies
investigating the effects of (real or presumed) preferential selection on
actual performance (Brown, Charnangavej, Keough, Newman, & Pentfrow,
2000; Turner & Pratkanis, 1993). Of particular relevance to our argument is
the study reported by Brown et al. investigating the performance of students
at the University of Texas belonging to academically stigmatised (Black and
Latino) or non-stigmatised groups (White and Asian). The authors found
that students belonging to stigmatised groups showed greater suspicion that
they may have received preferential treatment at admission to the university.
Compared to those belonging to non-stigmatised groups, they had also
obtained lower GPAs (US college grades) during their first year. Most
importantly, the effect of group membership on GPA was reliably mediated
by suspicion of preferential treatment. Thus, the mere suspicion of low-
status minority members that they may have been admitted to a university
(or job) on the basis of affirm ative action may be sufficient to impair their
performance. Assuming that affirm ative action is a potent reminder of
stereotypes associating low-status groups with low performance, Brown et
al’s findings seem consistent with a stereotype threat account, although they
are also open to different interpretations.
Along the same line, we have recently cond ucted a study (Carnaghi,
Maass, Benetti, & Callegari, 2003, Exp. 1), in which students were led to
believe that they could obtain a desirable short-term job.
3
Job selection was
based on a mathematics test, a domain in which women are assumed to do
poorly according to socially shared stereotypes. The hiring criterion was
varied so that it was either based on merit (the best candidate would be
hired) or on an affirmative action rule that either favoured men or women.
4
8
3
In reality, the money was distributed equally across participants after debriefing.
4
In both cases, the application of the affirmative action rule was justified on the basis of the
information that ‘‘the number of men (women) who had applied for this job was inferior to the
number of women (men) applicants and that it seemed appropriate to guarantee a reasonable
representation of men (women)’’
24 MAASS AND CADINU
We expected that the introduction of an affirmative action rule would
make gender highly salient. Indeed, regardless of whether men or women
were favoured by the affirmative action rule, a word completion task
indicated that gender-related concepts were much more accessible to
participants under affirmative action than under the merit rule. Also,
overall performance was much better under the merit than under the
affirmative action rule. This is not surprising considering that the hiring
decision is entirely contingent on the participant’s performance under the
merit rule, but only partiall y unde r the affirmative action rule. Thus,
participants should invest greater energy and effort into the task when the
decision is purely merit-based. More interesting is another prediction: If
affirmative action rules activate stereotypes, then, following Steele’s
model, women should perform particularly poorly under affirmative
action rules regardless of whether the rules favoured women or men. Our
data tend to confirm this prediction. Under affirmative action,
performance loss (compared to the merit condition) for men was 20%
when women were favoured by the affirmative action rules and 23%
when men were favoured. For women participants, the performance loss
was 38% when men were favoured and reached 47% when their own
group, women, were favoured.
If confirmed by future studies, these results suggest that explicit
affirmative action rules may activate existing stereotypes which, in turn,
may lead to a (stereotype-confirm ing) decline in performance among
minority members. The same minority member who would perform well
under merit rules may show considerable deficits under affirmative action.
Ironically, the very affirmative action programme that was developed to
protect low-status minority members against bias may harm their actual
performance.
To be perfectly clear, we do not suggest that affirmative action be
abolished. On the basis of the above results, we simply want to argue that
affirmative action rules may inadvertently activate existing stereotypes in the
minority’s mind and hence interfere with his or her performance. As Brown
et al.’s findings suggest, the mere suspicion of receiving preferential
treatment may be sufficient to undermine a minority person’s performance.
If confirmed by future studies, companies and universities may be well
advised to reassure minority members that merit (rather than group
membership) played a crucial role in their hiring or admission. Also, it may
be wise not to advertise affirmative action in close proximity to the selection
procedure since this may hinder (rather than facilitate) the minority’s
performance. We do believe that affirmative action is useful tool for
increasing minority participation in institutions in which they are under-
represented. At the same time, precautions should be taken to protect the
target of preferential treatment against the potentially damaging effects of
STEREOTYPE THREAT 25
stereotype activation that, unfortunately, are likely to accompany any
affirmative action rule.
PREVENTING STEREOTYPE THREAT
The ultimate goal of stereotype threat research is to identify strategies that
are able to prevent (or even invert) stereotype-related performance deficits.
To date, five different strategies have been investigated. The first, and most
straightforward method is to provide counter-stereotypical information
(Spencer et al., 1999). For examp le, in one of our studies we provided
information that women are doing better than men (Cadinu et al., 2003a).
The disadvantage of this method is that the information may at times not be
credible enough to override widely held beliefs.
The second strategy is to activate, within the stigmatised group, a
different category membership that is not negati vely linked to the task
domain. This may either be achieved by varying the comparative intergroup
context or by having people think about themselves in terms of different
category memberships. An example of the former strategy is to tell students
that a given maths test is administered with the explicit goal of comparing
American vs Canadians (rather than males vs females). This procedure
seems able to cancel the gender gap in maths performance (see Walsh,
Hickey, & Duffy, 1999). An excellent example of this strategy is also
provided by the work of Ambady, Shih and collaborators discussed earlier
(Ambady et al., 2001; Shih et al., 1999).
The third prevention strategy is to modify the test label so as to eliminate
any link to the category. For example, labelling a logical-mathematical test
as measuring ‘‘social intelligence’’ seems sufficient to reliably increase the
performance of women (see study by Cadinu et al., 2002b, described earlier).
A fourth strategy has recently been pro posed by Aronson, Fried, and
Good (2002). With reference to Dweck’s (1999) model, these authors argued
that teaching Black students to endorse an incremental rather than an entity
theory of intelligence woul d make them less vulnerable to stereotype threat.
In a very clever design, Black students were made to believe that they were
going to tutor younger students with academic problems and hence in need
of encouragement. Half of the students were instructed to convince their
younger pen pals that intelligence was malleable and may expand with hard
work (incremental theory). The remaining participants were also told to
encourage their pen pals, but this time the multifaceted nature of intelligence
was stressed rather than its malleability. Unbeknown to the parti cipants, the
manipulation served to change their own mind set. Results showed that
training students to adopt an incremental theory of intelligence reliably
improved their attitudes towards academics as well as their actual grades.
Thus, reshaping minority members’ implicit beliefs about intelligence may
6
26 MAASS AND CADINU
protect them against the debilitating effects of stereotype threat, without
modifying their perception of stereotype threat itself.
A fifth strategy that has produced pro mising results is the introduction of
a counter-stereotypical role model (Maass et al., 2002; Marx & Roman,
2002; McIntyre, Paulson, & Lord, 2003). Evidence for the potential role of a
counter-stereotypical role mod el as a buffer against stereotype threat comes
from a recent article by Marx and Roman (2002, Exps. 2 and 3). The
authors tested the hypothesis that learning about a competent female role
model would protect women against the debilitating effects of stereotype
threat on a difficult maths test. Indeed, in both studies female students
performed significantly better after learning about another, similar female
student with good (rather than poor) maths abiliti es.
In McIntyre et al.’s (2003) research, women participants learned about
other women’s achievements in domains such as architecture or medicine
that had no obvious link to mathematics. In the control condition,
participants read the same success stories, but this time they were attributed
to gender-unspecific companies. The performance of participants exposed to
successful female role models was reliably better than that of women who
had read the same success stories referred to companies. Thus, it is not the
success story per se, but the successful female role model that seems to
remove stereotype-related barriers.
We have recently extended this reasoning to a very different population,
name elderly people. In an as yet unpublished study by Maass et al. (2002),
we asked people over 65 years of age (whose age had been made salient) to
take a series of memory tests, including a text recall and an object
recognition test. In the stereotype threat baseline condition, participants
were simply told that these tests measured memory, informat ion that was
sufficient to activate the widely held idea of age-related memory deficits. Not
surprisingly, our participants performed very poorly in this condition. In the
second condition (counter-stereotypical information) they were told that,
for the particular type of memory investigated here, there was no age-related
decline and, indeed, elderly people often performed better than younger
ones. This manipulation strongly and reliably improved performance both
on recall and recognition. In the third, and most interesting condition, the
same instructions were provided as in the baseline condition, but this time a
counter-stereotypical role model was introduced. Participants were told that
the study was promot ed by the (fictitious) Levi-Montalcini Institute and a
recent photograph of the ‘‘Nobel Prize winner’’ Levi-Montalcini (at the age
of 92) appeared on the cover page of the test. Interestingly, the presence of
an elderly role model with exceptional intellectual abilities had an enormous,
facilitating effect on our participants. When Levi-Montalcini was introduced
as a role model, participants not only outperformed those in the baseline
condition, they even improved somewhat beyond the condition in which
STEREOTYPE THREAT 27
explicit counter-stereotypical information had been provided. A subsequent
study showed that the Nobel Prize winner could serve as a role model for
our elderly participants only if she was of high age; the model of a younger
Nobel Prize winner was unable to improve memory performance. Taken
together, these studies suggest that counter-stereotypical role models
provide an efficient and easily applicable strategy for reducing undesirable
stereotype threat effects.
Although not all of the methods outlined here may be equally
appropriate for all applied settings (e.g., it may not be very credible if a
maths teacher claims to assess social abilities in a test), we believe that they
are a promising starting point for the development of prevention strategies.
CONCLUSION
Only few years have passed since Steele and Aronson’s (1995) groundbreak-
ing work on stereotype threat, but during these few years a great number of
researchers around the world have been intrigued by this model and by its
applied implications. Minority members already face numerous barriers to
success including economic and cultural disadvantages and prejudiced
attitudes of teachers and employers that may bias performance evaluation,
hiring decisions etc. Steele and Aronson’s model suggests that, in addition to
these external barriers, members of stigmatised groups also seem to be
vulnerable to a much more subtle, but equally powerful impact of
stereotypes. In many test situations, minority members are disadvantaged
simply because their category membership is activated together with the
stereotypes that go along with it. The mere activation of stereotypes in the
test situation seems sufficient to produce a reliable interference, resulting in
greatly reduced performance. As our brief overview of the literature shows,
this idea has now received remarkab le support in very different paradigms,
with different minority populations, and in a wide variety of performance
domains. Although many issues, including the identification of the under-
lying processes, pose challenges for future research, the fast growing body of
literature (including our own work reported here) is largely supportive of the
basic predictions of the stereotype threat model. Even a short-lived
activation of task relevant stereotypes is sufficient to induce performance
deficits in those belonging to stigmatised social groups.
At first sight, the stereotype threat model seems to offer a very pessimistic
outlook, as it suggests that even the most subtle reminder of the minority’s
membership in a stigmatised group reduces its chances of success. But
fortunately the very same mod el can also be read in a positive light as it
offers insights into how to explain (apparent) differences in performance and
how to remove important barriers to perfor mance. Many social groups such
as women, Blacks, and the elderly have been found to perform poorly in
28
MAASS AND CADINU
specific domains such as chess and maths for women, verbal abilities for
Blacks, and memory tasks for the elderly. If Steele and Arons on’s model is
correct, then such under-performance is, at least in part, attributable to
stereotype threat rather than to deeper deficits in domain-relevant
capacities. Ultimately, it may be much easier to remove stereotype
activation from test situations (e.g., by eliminating gender or race
identification, by changing test labels, or by introducing successful role
models) than to compensate for profound deficits.
Clearly futur e research is faced with many difficult tasks. One is to
achieve a better understanding of the processes underlying stereotype threat
deficits, both for minority and majority members. Second, because not all
individuals may be equally vulnerable to the effects of stereotype threat, it is
crucial to identify those personal characteristics that may either protect or
put individuals at higher risk for stereotype-confirming performance. Third,
it is important to identify social contexts and situational constraints (such as
unbalanced numerical work settings) that are particularly conducive for
stereotype threat effects to emerge. Finally, we are left with the important
and most challenging task of proposing applied research and promoting
prevention programmes to weaken or override the detrimental effects of
stereotype threat.
REFERENCES
Ambady, N., Shih, M., Kim, A., & Pittinsky, T. L. (2001). Stereotype susceptibility in children:
Effects of identity activation on Quantitative performance. Psychological Science, 12, 385
390.
Aronson, J., 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.
Aronson, J., Lustina, M. J., Good, C., Keough, K., Steele, C. M., & Brown, J. (1999). When
white men can’t do math: Necessary and sufficient factors in stereotype threat. Journal of
Experimental Social Psychology, 35, 29 46.
Blascovich, J., Spencer, S. J., Quinn, D., & Steele, C. (2001). African Americans and high blood
pressure: The role of stereotype threat. Psychological Science, 12 , 225 229.
Brown, R. P., Charnangavej, T., Keough, K. A., Newman, M. L., & Pentfrow, P. J. (2000).
Putting the ‘‘affirm’’ into affirmative action: Preferential selection and academic
performance. Journal of Personality and Social Psychology, 79, 736 747.
Cadinu, M., Maass, A., Frigerio, S., Impagliazzo, L., & Latinotti, S. (2003a). Stereotype threat:
The effect of expectancy on performance. European Journal of Social Psychology, 33, 267
285.
Cadinu, M., Maass, A., & Lombardo, M. (2002). Stereotype threat: The moderating role of
individual differences. Manuscript submitted for publication.
Carnaghi, A., Maass, A., Benetti, A., & Callegari, L. (2003). Affirmative action: turning a pro-
minority policy into a barrier. Unpublished manuscript: Padova University.
Chacko, T. I. (1982). Women and equal employment opportunity: Some unintended effects.
Journal of Applied Psychology, 67, 119 123
9
8
STEREOTYPE THREAT 29
Cheryan, S., & Bodenhausen, G. V. (2000). When positive stereotypes threaten intellectual
performance: The psychological hazards of ‘‘Model Minority’’ status. Psychological Science,
11, 399 402.
Copeland, J. T. (1994). Prophecies of power: Motivational implications of social power for
behavioural confirmation. Journal of Personality and Social Psychology, 67, 264 277.
Crocker, J., & Major, B. (1989). Social stigma and self-esteem: The self-protective properties of
stigma. Psychological Review, 96, 608 630.
Croizet, J., & Claire, T. (1998). Extending the concept of stereotype and threat to social class:
The intellectual underperformance of students from low socio-economic background.
Personality and Social Psychology Bulletin, 24, 588 594.
Dembo, M. H., & Eaton, M. J. (1997). School learning and motivation. In G. D. Phye (Ed.),
Handbook of academic learning: Construction of knowledge. The educational psychology
series. (pp. 65 103). San Diego, CA: Academic Press.
Dion, K. L. (1975). Women’s reactions to discrimination from members of the same or opposite
sex. Journal of Research in Personality, 9, 294 306.
Dweck, C. S. (1999). Self-theories: their role in motivation, personality and development.
Philadelphia, PA: Taylor and Francis.
Eccles, J. S., Wigfield, A., Harold, R. D., & Blumenfeld, P. (1993). Age and gender differences in
children’s self- and task perceptions during elementary school. Child-Development, 64, 830
847.
Eccles, J. S., Wigfield, A., & Schiefele, U. (1998). Motivation to succeed. In W. Damon (Series
Ed.) & N. Eisenberg (Vol. Ed.), Handbook of child psychology (5th Ed., Vol. III, pp. 1017
1095). New York: Wiley.
Fagot, B. I., & Leinbach, M. D. (1993). Gender-role development in young children: From
discrimination to labeling. Developmental-Review, 1993, 205 224.
Good, C. D. (2001). The development of stereotype threat and its relation to theories of
intelligence: Effects on elementary school girls’ mathematics achievement and task choices.
Unpublished Doctoral Thesis, University of Texas at Austin.
Heilman, M. E., Block, C. J., & Lucas, J. A. (1992). Presumed incompetent? Stigmatization and
affirmative action efforts. Journal of Applied Psychology, 77, 536 544.
Heilman, M. E., Lucas, J. A., & Kaplow, S. R. (1990). Self-derogating consequences of sex-
based preferential selection: The moderating role of initial self-confidence. Organizational
Behaviour and Human Decision Processes, 46, 202 216.
Hill, K. T., & Wigfield, A. (1984). Test anxiety: A major educational problem and what can we
do about it. Elementary School Journal, 85, 105 126.
Hilton, J. L., & Darley, J. M. (1985). Constructing other people: A limit on the effect. Journal of
Experimental Social Psychology, 21, 1 18.
Inzlicht, M., & Ben-Zeev, T. (2000). A threatening intellectual environment: Why females are
susceptible to experiencing problem-solving deficits in the presence of males. Psychological
Science, 11, 365 371
Jussim, L. (1986). Self-fulfilling prophecies: A theoretical and integrative review. Psychological
Review, 93, 429 445.
Keller, C. (2001). Effect of teachers’ stereotyping on students’ stereotyping of mathematics as a
male domain. Journal of Social Psychology, 141, 165 173.
Killen, M., Pisacane, J., Lee-Kim, J, & Ardila-Rey, A. (2001). Fairness or stereotypes? Young
children’s priorities when evaluating group exclusion and inclusion. Developmental
Psychology, 37, 587 596.
Leyens, J.-P., De
´
sert, M., Croizet, J.-C., & Darcis, C. (2000). Stereotype threat: Are lower status
and history of stigmatization preconditions of stereotype threat? Personality and Social
Psychology Bulletin, 26, 1189 1199.
3
30 MAASS AND CADINU
Maass, A., Cadinu, M., Verga, S., & Crimaldi, S. (2002). Preventing stereotype-induced memory
deficits among the elderly: The Levi-Montalcini effect (unpublished manuscript).
Martin, C. L. (1993). New directions for investigating children’s gender knowledge.
Developmental Review, 13, 184 204.
Marx, T., & Roman, J. S. (2002). Female role models: Protecting women’s math test
performance. Personality and Social Psychology Bulletin, 28, 1183 1193.
McIntyre, R. B., Paulson, R. M., & Lord, C. G. (2003). Alleviating women’s mathematics
stereptype threat through salience of group achievements. Journal of Experimental Social
Psychology, 39, 83 90.
Merton, R. K. (1948). The self-fulfilling prophecy. Antioch Review, 8, 192 210.
Mezzapesa (1999 2000). La minaccia legata allo stereotipo: Effetti su un campione di meridionali
[Stereotype threat: effects on a Southern-Italian sample]. Unpublished Laurea Thesis,
University of Padova.
Osborne, J. W. (2001). Testing stereotype threat: Does anxiety explain race and sex differences
in achievement? Contemporary Educational Psychology, 26, 291 310.
Pinel, E. C. (2002). Stigma consciousness in intergroup contexts: the power of conviction.
Journal of Experimental Social Psychology, 38, 178 185.
Pintrich, P. R., Marx, R. W, & Boyle, R. A. (1993). Beyond cold conceptual change: The role of
motivational beliefs and classroom contextual factors in the process of conceptual change.
Review of Educational Research, 63, 167 199.
Rosenthal, R. (1994). Interpersonal expectancy effects: a 30-year perspective. Current Directions
in Psychological Science, 3, 176 179.
Rosenthal, R., & Jacobson, L. (1968). Pygmalion in the classroom. New York: Holt, Rinehart &
Winston.
Schmader, T. (2002). Gender identification moderates stereotype threat effects on women’s
math performance. Journal of Experimental Social Psychology, 38, 194 201.
Sekaquaptewa, D., & Thompson, M. (2003). Solo status, stereotype threat, and performance
expectancies: Their effects on women’s performance. Journal of Experimental Social
Psychology, 39, 68 74.
Shih, M., Pittinsky, T. L., & Ambady, N. (1999). Stereotype susceptibility: Identity salience and
shifts in quantitative performance. Psychological-Science, 10, 80 83.
Signorella, M. L., Bigler, R. S., & Liben, L. S. (1993). Developmental diffferences in children’s
gender schemata about others: a meta-analytic review. Developmental Review, 13, 147 183.
Snyder, M., & Haugen, J. A. (1995). Why does behavioural confirmation occur? A functional
perspective on the role of the target. Personality and Social Psychology Bulletin, 21, 963
974.
Snyder, M., & Stukas, A. A. (1999). Interpersonal processes: The interplay of cognitive,
motivational, and behavioural activities in social interaction. Annual Review of Psychology,
50, 273 303.
Spencer, S. J., Steele, C. M., & Quinn, D. M. (1999). Stereotype threat and women’s math
performance. Journal of Experimental Social Psychology, 35, 4 28.
Stangor, C., Carr, C., & Kiang, L. (1998). Activating stereotypes undermines task performance
expectations. Journal of Personality And Social Psychology, 75, 1191 1187.
Steele, C. M., & Aronson, J. (1995). Stereotype threat and the intellectual test performance of
African Americans. Journal of Personality and Social Psychology, 69, 797 811.
Stone, J., Lynch, C. I., Sjomeling, M., Darley, J. M. (1999). Stereotype threat effects on Black
and White athletic performance. Journal of Personality and Social Psychology, 77, 1213
1227.
Stukas, A. A., & Snyder, M. (2002). Target’s awareness of expectations and behavioural
confirmation in ongoing interactions. Journal of Experimental Social Psychology, 38, 31 40.
AT
2
STEREOTYPE THREAT 31
Swim, J. K., & Stangor, C. (1998). Prejudice: The target’s perspective. San Diego, CA: Academic
Press.
Tiedemann, J. (2000). Parents’ gender stereotypes and teachers’ beliefs as predictors of
children’s concept of their mathematical ability in elementary school. Journal of Educational
Psychology, 92, 144 151.
Thompson, S. K. (1975). Gender labels and early sex role development. Child Development, 46,
339 347.
Turner, M. E., & Pratkanis, A. R. (1993). Effects of preferential and meritorious selection on
performance: An intuitive and self-handicapping perspective. Personality and Social
Psychology Bulletin, 19, 47 58.
Valian, V. (1998). Why so slow? The advancement of women. Cambridge, MA: MIT Press.
Vorauer, J. D., & Miller, D. T. (1997). Failure to recognize the effect of implicit social influence
on the presentation of the self. Journal of Personality and Social Psychology, 73, 281 295.
Walsh, M., Hickey, C., & Duffy, J. (1999). Influence of item content and stereotype situation on
gender differences in mathematical problem solving. Sex Roles, 41, 219 40.
Wigfield, A. (1994). Expectancy-value theory of achievement motivation: A developmental
perspective. Educational Psychology Review, 6, 49 78.
Wigfield , A., & Eccles, J. S. (1992). The development of achievement task values: A theoretical
analysis. Developmental Review, 12, 265 310.
Wigfield, A., & Eccles, J. S. (2000). Expectancy-value theory of achievement motivation.
Contemporary Educational Psychology, 25, 68 81.
Wigfield, A., Eccles, J. S., Yoon, K. S., Harold, R. D., Arbreton, A., & Freedman-Doan, K. et
al. (1997). Changes in children’s competence beliefs and subjective task values across
elementary school years: A three-year study. Journal of Educational Psychology, 89, 451
469.
32 MAASS AND CADINU
European Review of Social Psychology
Typeset by Elite Typesetting for
PERS
Manuscript No.
Author
Editor
Master
Publisher
QUERIES: to be answered by AUTHOR
AUTHOR:
The following queries have arisen during the editing of your manuscript.
Please answer the queries by marking the requisite corrections at the appropriate positions
in the text.
QUERY NO.
QUERY DETAILS
QUERY ANSWERED
www.elitetypesetting.com
T
Y
P
E
S
E
T
T
I
N
G
T
E
C
H
N
I
Q
U
E
S
L
T
D
0017
1 Allport, 1954 not listed in references
2 Spelling? Sekaquapteka in text, Sekaquaptewa in references
3 Cheryan & Bodenhausen, 1999 is listed as 2000 – which is correct?
4 Spencer et al., 1998 not listed
5 Maass, Cadinu & Verga, 2002 not listed. Is this Maass, Cadinu,
Verga & Crimaldi, 2002?
6 Cadinu et al., 2003b not listed
7 Svenson & Maule, not listed
8 Carnaghi et al., 2003 in prep update if possible
9 Cadinu et al., 2002 submitted update if possible and amend all
text citations to match
10 Is this manuscript available from e.g. a university
... Trend viših akademskih postignuća kod devojčica registruje se godinama unazad i u zemljama zapada (EACEA/Eurydice, 2010;European Comission, 2013;Sadker & Sadker, 2010). Dečaci u Srbiji su znatno češće uključeni u inkluzivno obrazovanje -dve trećine dece sa individualnim obrazovnim planom su dečaci (Republički zavod za statistiku, 2017), a školski neuspeh i ponavljanje razreda je i u našoj sredini problem s kojim se nešto češće suočavaju dečaci (Jovanović, Čekić Marković, Veselinović, Vušurović, i Jokić, 2017;Malinić, 2009). ...
... Slika 1. šematizuje ovaj mehanizam. Delovanje ovog mehanizma prvi put je pokazano na primeru rasnih stereotipa, a zatim je pokazano i da aktivacija rodnih, etničkih i drugih stereotipa takođe može negativno uticati na postignuće u testovnim situacijama (za pregled i meta-analizu pogledati Maass & Cadinu, 2003;Nguyen & Ryan, 2008). Efekti pretnje stereotipom demonstrirani su već na predškolskom uzrastu (Ambady, Shih, Kim, & Pittinsky, 2001: Tomasetto, Alparone, & Cadinu, 2011, kao i na starijim školskim uzrastima i u odraslom dobu (Keller, 2007;Keller & Dauenheimer, 2003;Maass & Cadinu, 2003;Nguyen & Ryan, 2008). ...
... Delovanje ovog mehanizma prvi put je pokazano na primeru rasnih stereotipa, a zatim je pokazano i da aktivacija rodnih, etničkih i drugih stereotipa takođe može negativno uticati na postignuće u testovnim situacijama (za pregled i meta-analizu pogledati Maass & Cadinu, 2003;Nguyen & Ryan, 2008). Efekti pretnje stereotipom demonstrirani su već na predškolskom uzrastu (Ambady, Shih, Kim, & Pittinsky, 2001: Tomasetto, Alparone, & Cadinu, 2011, kao i na starijim školskim uzrastima i u odraslom dobu (Keller, 2007;Keller & Dauenheimer, 2003;Maass & Cadinu, 2003;Nguyen & Ryan, 2008). Slika 1. Šematski prikaz mehanizma Pretnja stereotipom (Steele & Aronson, 1995) Kada je reč o rodnom stereotipu o matematičkim sposobnostima, podsećanje žena na sadržaj ovog stereotipa dovešće do pada njihovog postignuća na testu iz matematike u odnosu na situaciju u kojoj ne postoji ovakvo podsećanje. ...
Thesis
Full-text available
Even though academic gender stereotypes have been identified as a key psychological barrier to achieving full potential of women in traditionally masculine and men in traditionally feminine academic-career domains, supportive empirical evidence so far has been mixed, while findings on mechanisms through which stereotypes could influence achievement and academic-career choices have been scarce. This study set out to explore whether, to what extent and through which mechanisms implicit and explicit math and language gender stereotypes in primary school teachers and their students contribute to gender differences in educational outcomes. We tested three hypothesized mediating mechanisms to this relation: at the intrapersonal level, students’ mathematical and linguistic self concept; at the interpersonal level, teachers’ expectations from students and their gender differential treatment in the classroom. In a nested design, during the first phase of the study, we measured gender academic stereotypes of 115 primary school teachers, along with their academic expectations from 2295 students, and these students’ grades. In the second phase, we subsampled 16 classes comprising of 412 students, and measured their gender academic stereotypes, academic self-concept and test achievement. In addition, we observed the dyad interaction between the teachers and students in a total of 56 mathematics and Serbian language classes. We observed that educational outcomes can be predicted based on the explicit gender academic stereotypes of the students, albeit with small predictive and only in domains where the superiority of a specific gender would be expected according to gender stereotype (boys’ achievement and aspirations in the domain of mathematics and girls’ aspirations in the domain of language). These effects were mediated via academic self-concept in the corresponding domain, although a direct effect of gender stereotypes was still observed. The effects of teachers' gender academic stereotypes were also small and moderated by the students’ gender. Explicit stereotypes of teachers negatively affected the expectations and assessments, but not knowledge of girls in both academic domains, and positively affected the knowledge and aspirations of boys in the field of mathematics. Although teachers exhibited gender-biased expectations and gender-differential treatment of students in the classroom, we found no evidence of them mediating the relation between teachers’ gender stereotypes and students’ educational outcomes. We related the findings to similar research and discussed them in Balanced Identity Theory framework. Finally, we articulated guidelines for future research and educational policies.
... Although the research has typically focused on negative stereotypes, Cheryan et al. (2000) have reported evidence that even positive stereotypes might cause underperformance in specific tasks at hand. Beyond these immediate performance effects, substantial evidence also suggests a more long-term dis-identification (decreased desire to participate in and ultimately withdrawal) from the tasks and even negative physical health consequences (Maass et al., 2003;Spencer et al., 2016). Such effects have been observed in many stereotyped groups 4 and in relation to both intellectual and physical tasks. ...
... Such effects have been observed in many stereotyped groups 4 and in relation to both intellectual and physical tasks. A variety of mechanisms ranging from working memory depletion through to over-thinking automated processes have been proposed (Maass et al., 2003;Schmader et al., 2008;Spencer et al., 2016). The frightening implication of stereotype threat is that, even if the stereotypes held by an in-group are suppressed to the extent that they do not find expression in affective or behavioural responses, they can still have negative consequences that can be self-fulfilling within the out-group. ...
... Woodcock et al. (2012) reason that stereotype threat is a dynamic process and its effects consequently influence an individual's cognition and motivation (Bedyńska et al., 2020). The underperformance of the stereotyped target can be seen as a general phenomenon when the group's negative stereotype is made salient (Maass and Cadinu, 2003 communication; avoid direct social interactions as they find such encounters more stressprovoking and resource-depleting. Mowrer's (1951) two-factor theory of avoidance offers some guidance on how and why people may opt to remove fear-provoking stimuli to eliminate or diminish unpleasant emotions, thereby reinforcing their avoidance behavior. ...
Article
Purpose – Language plays an important role in a successful service exchange, but it can become a source of discrimination if one party is a non-native speaker in the host country. This study examines the linguistic racism that non-native customers experience in Inter Culture Service Encounters (ICSEs) and delve into factors that contribute to the underlying psychological responses and the behavioral outcomes. Design/methodology/approach – A phenomenological approach was used where 16 individuals were interviewed to discover themes through non-native customers' lens using an inductive process. Next, the emerged categories were classified based on extant literature, using a deductive approach. Findings – The findings highlight the role of language varieties as a strong social identity cue for non-native customers where the associated stigma makes them see ICSE as a stereotype threat. Most importantly, these experiences shape their future behavior by avoiding direct interactions with the servers and adopting other service channels. Several ‘social others’ influence this process. Originality/value – This study explores the notion of linguistic racism in an ICSE from a non-native consumers’ lens and thus adds to this under-researched literature. Using a phenomenological approach, we propose a framework focusing on the perception of language-related stigma and discrimination experienced by non-native consumers along with possible behavioral responses.
... Ενώ η προσδοκία της επιτυχίας είναι σημαντικός παράγοντας που διαμεσολαβεί στην αξία που δίνεται στη δραστηριότητα, ο διαμεσολαβητικός ρόλος των στερεοτύπων είναι διττός. Μπορεί είτε να διευκολύνει είτε να αποτρέψει αυτήν τη διαμεσολάβηση (Maass & Cadinu, 2003). Αυτό θα επέτρεπε ένα άτομο το οποίο είναι μέλος μιας ομάδας, για την οποία υπάρχουν αρνητικά στερεότυπα, να κάνει μια συνειδητή προσπάθεια προκειμένου να διαψεύσει τα στερεότυπα αυτά. ...
Article
Στην παρούσα έρευνα επιχειρείται η κατανόηση των δυσκολιών που έχουν τα παιδιά Ρομά στο σχολείο μέσα από τη διερεύνηση της διαδικασίας της αυτο-κατηγοριοποίησής τους. Ελέγχθηκε κατά πόσο οι στερεοτυπικές θέσεις της κυρίαρχης ομάδας για τους Ρομά υιοθετούνται από τα ίδια τα παιδιά μέσα από διαδικασία της αυτο-κατηγοριοποίησης. Ειδικότερα, διερευνώνται οι στερεοτυπικές θέσεις της ελληνικής κοινωνίας, που αφορούν στη σχολική συμπεριφορά, στη σχολική επίδοση και στην επαγγελματική προοπτική των Ρομά. Στην έρευνα συμμετείχαν 40 νήπια και παιδιά Ρομά σχολικής ηλικίας. Μέσα από την ταύτισή τους με ηρωίδα βιντεοσκοπημένου κουκλοθεάτρου, η οποία βρίσκονταν σε διομαδική σύγκριση, τα παιδιά απάντησαν για τις παραπάνω διαστάσεις της σχολικής ζωής. Τα αποτελέσματα της έρευνας έδειξαν ότι οι στερεοτυπικές θέσεις της έξω ομάδας αναγνωρίζονται από τους συμμετέχοντες ως χαρακτηριστικά της ταυτότητάς τους πριν ακόμη έρθουν σε επαφή με το σχολείο. Οι συμμετέχοντες πριν τη σχολική φοίτηση θεωρούν ότι τα παιδιά Ρομά είναι αυτά που έχουν προβλήματα συμπεριφοράς στο σχολείο, δυσκολίες στα μαθήματα, και θα επιλέξουν επαγγέλματα χαμηλού κοινωνικού κύρους. Η επαφή με τις πραγματικές διομαδικές συνθήκες κατά τη σχολική φοίτηση, ωστόσο, αλλάζουν κάποια από τα στερεότυπα που αποκτήθηκαν μέσα από τη διαδικασία της αυτο-κατηγοριοποίησης. Προτείνεται αξιοποίηση των συμπερασμάτων στα προγράμματα παρέμβασης.
... Evidence shows that when facing stereotype threat during a task, the ventral anterior cingulate cortex-a brain area associated with the control of emotions and the process of social feedback-is activated, leading to a drop in performance (Wraga et al., 2007;Krendl et al., 2008). The processes through which stereotype threat lead to a decrease in performance thus arise from increased anxiety, disturbing thoughts, overly cautious tendencies, pessimism, and disengagement (Maass and Cadinu, 2003). Anyone may experience a degree of anxiety while performing a task, but stereotype threat is known to place an additional load on members of the stereotyped groups. ...
Article
Full-text available
In this study we aim to analyze the combined effect of age-based and gender stereotype threat on work identity processes (and in particular on authenticity and organizational identification) and on work performance (self-rating performance). The research utilizes an ample sample of over fifty-year-old workers from diverse organizations in Italy. Using a person-centered approach four clusters of workers were identified: low in both age-based and gender stereotype threat (N = 4,689), high in gender and low in age-based stereotype threat (N = 1,735), high in age-based and low in gender stereotype threat (N = 2,013) and high in both gender and age-based stereotype threat (N = 758). Gender was significantly associated with these clusters and women were more frequently present in those groups with high gender stereotype threat. ANOVA results show that workers in the last two clusters score significantly lower in authenticity, organizational identification and self-rate performance. All in all, if ageism is undoubtedly problematic for older workers' identity processes, ageism and gender-stereotypes represent a double risk for women over fifty in the workplace. The analysis of the results can be beneficial both for the theoretical advancement and for the practical insights offered in the organizational and management field, where new policies of HR management can be elaborated, in order to value and to improve the workers experience.
... The opposite was also true, if told that they are expected to be superior they performed so much better. (Maass & Cadinu, 2003;Hoff & Pandey, 2006). ...
Article
In 2015, a video of policemen beating up an Ethiopian-born first Lieutenant in the Israeli Defense Forces, was published by the Israeli media and triggered a massive protest against police brutality and discrimination of the Ethiopian community. The current study aimed to understand the meaning the members of the Ethiopian community attribute to the protest, and its' relation to their experiences within the Israeli society. The paper is based on data gathered through interviews with 19 young Ethiopian Israeli adults. The analysis revealed that the participants' interaction with Israeli society is characterized by a shared experience of discrimination and racism, which shaped their perception of protest as a means of speaking out, strengthening a collective identity and achieving feelings of empowerment. However, individual differences were found in the way the participants believed their protest should be conducted. Findings suggest that an understanding of protest should take into account not only shared group experience but also illuminate individual differences. Universalist understandings of protest should be widened to include an examination of how protest can inform us about the social and historical process of relations between a minority group and the majority group.
Article
This study contrasted the effects of two task messages, evaluative or non-evaluative, on mathematics performance, affect, and intrinsic task motivation. One hundred-twenty secondary-school students aged 17–21 years were delivered one of the two messages, or assigned to a control condition, before completing a mathematics task, measures of message appraisals (challenge and threat), affect (pleasantness, arousal, dominance), and a behavioural indication of intrinsic task motivation. The evaluative message raised performance only in males, while for females both messages decreased intrinsic motivation for the task, probably due to stereotype threat. Implications for future research and educational practices are discussed. • HIGHLIGHTS • In a low-value context, an evaluative message favoured male mathematics performance • Males increased arousal after an evaluative message • A challenge appraisal was linked with male performance • Females decreased intrinsic motivation after evaluative and non-evaluative messages
Article
Full-text available
Seventy-two children (35 4½-year-olds and 37 5½-year-olds), nearly evenly divided by gender, from European American (71%), Asian American (17%), and African American (12%) middle-class backgrounds, were individually interviewed about straightforward exclusion and inclusion for two gender-stereotypic peer-group contexts: activities (doll and truck play) and role-play (teacher and firefighter). All children evaluated straightforward exclusion based on gender (e.g., girls excluding a boy from doll play) as wrong and used moral reasons. Preliminary inclusion decisions in the activity contexts (choosing a boy or a girl to join the group) were based on stereotypic expectations, particularly for younger children. Given the opportunity to weigh alternative considerations, however, all children gave priority to fairness over stereotypic expectations in both multifaceted inclusion peer-group contexts.
Article
Full-text available
Recent theory and research suggest that certain situational factors can harm women’s math test performance. The three studies presented here indicate that female role models can buffer women’s math test performance from the debilitating effects of these situational factors. In Study 1, women’s math test performance was protected when a competent female experimenter (i.e., a female role model) administered the test. Study 2 showed that it was the perception of the female experimenter’s math competence, not her physical presence, that safeguarded the math test performance of women. Study 3 revealed that learning about a competent female experimenter buffered women’s self-appraised math ability, which in turn led to successful performance on a challenging math test.
Book
Some years ago we, the editors of this volume, found out about each other's deeply rooted interest in the concept of time, the usage of time, and the effects of shortage of time on human thought and behavior. Since then we have fostered the idea of bringing together different perspectives in this area. We are now, there­ fore, very content that our idea has materialized in the present volume. There is both anecdotal and empirical evidence to suggest that time con­ straints may affect behavior. Managers and other professional decision makers frequently identify time pressure as a major constraint on their behavior (Isen­ berg, 1984). Chamberlain and Zika (1990) provide empirical support for this view, showing that complaints of insufficient time are the most frequently report­ ed everyday minor stressors or hassles for all groups of people except the elderly. Similarly, studies in occupational settings have identified time pressure as one of the central components of workload (Derrich, 1988; O'Donnel & Eggemeier, 1986).
Article
Two studies investigated whether a stigma of incompetence marks those associated with affirmative action programs. In an experiment, 129 male and female undergraduates reviewed the application materials of someone said to be recently hired for one of two jobs. The hiree was either a man or a woman, and the woman either was or was not associated with an affirmative action program. The affirmative action label was found to negatively affect the perceived competence of women hirees regardless of the degree to which the job was male sex-typed. A field investigation in which 184 White men provided information about their co-workers supported these results. It additionally demonstrated that the relationship between perceived competence and presumed affirmative action status held not only when co-workers were White women but when they were Black men and Black women as well. The affirmative action label also was associated with negative characterizations of activity and potency and, in the field study, interpersonal attributes and prognoses for career progress.
Chapter
This chapter discusses various aspects of school learning and motivation. Motivation is a topic introduced early in any discussion with teachers about student learning. Teachers espouse hypotheses or beliefs about student motivation that influence their interaction with students and the decisions that determine how they relate to students. Perspectives on motivation, social and cultural factors, classroom contextual factors, internal factors, teacher-directed interventions, authority dimension, recognition dimension, grouping dimension, evaluation dimension, time dimension, student self-regulation strategies, goal setting, and so on are discussed. “Socially negotiated” specifies that many rules of behavior or norms are developed by individuals in response to the needs of the environments in which they interact. Teachers, structure activities, reward, and interact with students in ways that influence the students' personal learning goals or intentions for performing a task. A performance goal focuses on social comparison and competition, with the main purpose of outperforming others on the task. Tasks that are interesting, meaningful, challenging, and authentic in terms of actual experiences—relevant to life outside school—can facilitate a mastery learning orientation.
Article
Inconsistencies that have been reported in past research on developmental changes in gender schemata actually may be a consequence of differences in the way these schemata have been conceptualized and measured. Meta-analysis was used to evaluate this interpretation of past work. On forced choice measures, in which children must select one sex or the other for each item (e.g., "Who is the strong one?"), "correct" matches to societal stereotypes increased with age. Increases were not, however, related to the type of question used (e.g., "Who is ...?" versus "Who can ...?"). Girls made more stereotype matches than boys, although the magnitude of the effect was small. In contrast, on nonforced choice measures, type of question did affect results. Children showed increases in nonstereotyped responses with age, but especially when asked "Who should . . ." or "Who can . . ." , and when elementary-school-aged (as well as preschool-aged) children were included. Girls gave significantly more nonstereotyped responses than boys, especially among older samples and when the domain was traits. Both the age and the sex effects in nonstereotyped responses were larger in more recent studies. IQ and television viewing were significantly related to forced choice scores, whereas television viewing, maternal employment, and memory for gender-stereotyped material were all significantly related to non-forced choice scores. Implications for the distinction between knowledge of stereotypes and attitudes toward stereotypes are discussed.