Você está na página 1de 14

Child Development, January/February 2014, Volume 85, Number 1, Pages 250263

The Roots of Stereotype Threat: When Automatic Associations Disrupt Girls


Math Performance
Silvia Galdi and Mara Cadinu Carlo Tomasetto
University of Padova University of Bologna

Although stereotype awareness is a prerequisite for stereotype threat effects (Steele & Aronson, 1995), research
showed girls decit under stereotype threat before the emergence of mathgender stereotype awareness, and
in the absence of stereotype endorsement. In a study including 240 six-year-old children, this paradox was
addressed by testing whether automatic associations trigger stereotype threat in young girls. Whereas no indi-
cators were found that children endorsed the mathgender stereotype, girls, but not boys, showed automatic
associations consistent with the stereotype. Moreover, results showed that girls automatic associations varied
as a function of a manipulation regarding the stereotype content. Importantly, girls math performance
decreased in a stereotype-consistent, relative to a stereotype-inconsistent, condition and automatic associations
mediated the relation between stereotype threat and performance.

Stereotype threat research (Steele & Aronson, 1995) from self-reports, described as explicit measures,
has shown that both adults and children underper- which assess conscious beliefs.
form in difcult tests when a negative domain- This study employed implicit and explicit
relevant in-group stereotype is made salient (see measures jointly to show that automatic associa-
Inzlicht & Schmader, 2012, for a review). Despite tions consistent with a negative in-group stereotype
this overwhelming evidence, the specic require- may lead to stereotype-induced underperformance
ments for childrens underperformance under ste- at early stages of development, even though
reotype threat are still unclear. With the aim of children do not possess the cognitive competencies
providing new insights into research on stereotype for stereotype awareness yet, and in the absence of
threat, this study investigates automatic associations evidence that they endorse the stereotype content.
as a potential requirement for the emergence of ste-
reotype threat underperformance in children.
Prerequisites for Stereotype Threat Susceptibility
Automatic associations are those associations
between concepts (e.g., mewoman), or between According to the stereotype threat model, for ste-
concepts and attributes (e.g., owerpositive) that reotypes to affect performance, children need to (a)
come to mind unintentionally when an individual have developed a concept of social categories (cate-
encounters a relevant object, that are difcult to con- gory awareness), (b) be able to identify themselves as
trol once they have been activated, and that are not members of a social category (self-categorization), and
necessarily explicitly endorsed (e.g., Gawronski & (c) know that the in-group category is negatively
Bodenhausen, 2006). Such automatic associations are related to specic domains or attributes (stereotype
often contrasted with conscious beliefs, which are awareness; Aronson & Good, 2003). When children
those mental contents that an individual explicitly enter elementary school, they possess the cognitive
endorses as accurate (e.g., Gawronski & Bodenhau- competencies of category awareness and self-categoriza-
sen, 2006). Most measures of automatic associations tion (e.g., Martin & Ruble, 2010) but not stereotype
are based on participants performance on com- awareness (e.g., McKown & Strambler, 2009).
puter-based, speeded categorization tasks (see Gaw- Research has shown that, between the ages of 3
ronski & Payne, 2010, for a review). These measures, and 4 years, children become aware of social
commonly referred to as implicit measures, differ categories, such as gender and race (e.g., Aboud,

Correspondence concerning this article should be addressed to 2013 The Authors


Silvia Galdi, Dipartimento di Psicologia dello Sviluppo e della Child Development 2013 Society for Research in Child Development, Inc.
Socializzazione, Universita di Padova, Via Venezia 15, Padova All rights reserved. 0009-3920/2014/8501-0017
35131, Italy. Electronic mail may be sent to silvia.galdi@unipd.it. DOI: 10.1111/cdev.12128
Stereotype Threat Before Stereotype Awareness 251

1988; Katz & Kofkin, 1997), and are able to identify math performance among 5- to 7-year-old Italian
themselves as members of these categories (e.g., girls, and Neuville and Croizet (2007) obtained
Martin & Ruble, 2010; Quintana, 1998). Moreover, similar ndings among 7-year-old French girls.
by this age children develop personal stereotypic The last three studies raise a theoretical paradox:
beliefs about abilities and characteristics of ethnic How can stereotype-induced performance decre-
(Aboud, 1988) and gender groups. For example, ments be found in girls who do not possess stereo-
they hold personal stereotypic beliefs about gender type awareness yet? One could assume that for a
differences in toy preferences, dressing, and aggres- negative stereotype to affect performance in chil-
sive versus prosocial behaviors (Martin & Ruble, dren who have not developed stereotype awareness
2010). However, personal stereotypic beliefs, yet, it is sufcient that children hold the personal
referred to as stereotype endorsement, differ from the stereotypic belief that their in-group is less skillful
knowledge of stereotypes held by others (not neces- in the relevant domain. Indeed, given that young
sarily personally endorsed), which is dened as children do not distinguish their own stereotypic
stereotype awareness (e.g., Martinot & Desert, 2007; beliefs from others, and rather assume that their
McKown & Strambler, 2009). Thus, although in beliefs are shared by others as well (e.g., Augousti-
many cases personal beliefs overlap with socially nos & Rosewarne, 2001), in principle childrens
shared stereotypes, this does not imply necessarily personal negative beliefs about abilities of their
that young children are aware of stereotypes. in-group could be sufcient to trigger stereotype
Indeed, the literature on childrens social perspec- threat. If this were the case, stereotype endorsement
tive taking (e.g., Selman, 1980), theory of mind would be the key to identify the sources of stereo-
(e.g., Perner & Wimmer, 1985), and person percep- type threat in young children.
tion (e.g., Rholes & Ruble, 1984) suggests that chil- To date, the potential role of childrens endorse-
dren develop the cognitive competencies for ment of the mathgender stereotype as a prerequi-
stereotype awareness only by middle childhood, site for stereotype threat effects does not t with
and that up to then they do not distinguish their the available evidence. For example, Ambady et al.
personal stereotypic beliefs from others beliefs. (2001) found that 5- to 7-year-old American girls
Consistently, children have been shown to be aware believe that boys and girls are equal at math (for
of common ethnic and gender stereotypes about similar results on Italian children, see Tomasetto
academic abilities by the ages of 8 and 9 years (e.g., et al., 2011); on the contrary, 5- to 7-year-old Amer-
McKown & Weinstein, 2003; Quintana, 1998). ican boys state that boys are better at math, thus
The evidence that children do not possess stereo- exhibiting in-group favoritism (e.g., Powlishta,
type awareness when they enter elementary school 1995; Yee & Brown, 1994). Similarly, other research
has important consequences. As noted earlier, showed no endorsement of the mathgender stereo-
stereotype awareness is a specic requirement for type until 89 years of age among Italian and
stereotype threat. Specically, the stereotype threat French children (Martinot & Desert, 2007; Muzzatti
model posits that peoples negative stereotypes & Agnoli, 2007).
hamper performance only in situations that induce To our knowledge, only one study found stereo-
targets to become concerned about being judged by type endorsement as early as 67 years of age in a
others on the basis of relevant stereotypes (Steele, Western country (i.e., United States; Cvencek,
1997). Consistently, research on performance decre- Meltzoff, & Greenwald, 2011). However, this dis-
ments induced by ethnic stereotypes conrms that crepancy from the other ndings might be due to
only children who are aware of broadly held stereo- the type of measures used. Whereas all the studies
types are vulnerable to stereotype threat effects discussed elsewhere in the article assessed the
(McKown & Strambler, 2009; McKown & Weinstein, endorsement of the gender stereotype regarding
2003). Therefore, in principle we should not expect math ability, Cvencek et al. (2011) assessed the
declines in performance under stereotype threat endorsement of the stereotype that boys like math
prior to 89 years of age, or even later (Aronson & and girls like language. Therefore, it is possible that
Good, 2003). Nevertheless, Ambady, Shih, Kim, and 6- to 7-year-olds already hold the personal stereo-
Pittinsky (2001) showed that 5- to 7-year-old Asian type that boys like math, but still do not endorse the
American girls underperformed on a math task stereotype about girls and boys math ability.
when their gender identity was made salient, as Consistent with this reasoning, Steele (2003) found
compared to children in a control condition. that 6- to 10-year-old children do not endorse the
Similarly, Tomasetto, Alparone, and Cadinu (2011) stereotype regarding boys and girls abilities in math
found that gender identity activation hampered (see also Martinot, Bages, & Desert, 2012), even
252 Galdi, Cadinu, and Tomasetto

though they endorse the same stereotype about men be automatically activated (e.g., Rudman, 2004;
and women (stereotype stratication). Thus, to date, Wilson, Lindsey, & Schooler, 2000), whereas impli-
there is no evidence that young children endorse cit measures detect only those highly overlearned
the mathgender ability stereotype, at least in associations between concepts whose activation has
Western countries. Nevertheless, research shows become automatic over the course of long-term
that it is specically the ability component of the experiences. According to this widespread assump-
mathgender stereotype that affects womens per- tion, in terms of emergence, conscious beliefs reect
formance, whereas other stereotypical beliefs (e.g., more recently formed representations, whereas
women are less interested in math) do not trigger automatic associations reect only the older repre-
stereotype threat effects (Thoman, White, Yam- sentations. However, an emerging body of research
awaki, & Koishi, 2008). suggests that the opposite may be true as well.
To summarize, young girls show math underper- Recent ndings showing experimentally induced
formance before being aware of the mathgender changes in implicit but not explicit measures (e.g.,
stereotype, and in the absence of evidence that they Gawronski & LeBel, 2008; Olson & Fazio, 2006)
endorse such belief as a personal stereotype. One suggest that automatic associations may reect
possibility to explain this paradox is that previous more recently formed representations as compared
ndings on childrens mathgender stereotype have to the corresponding conscious beliefs. These results
been based only on self-reports, which may not are also consistent with studies using implicit and
capture all relevant aspects of the stereotype. Thus, explicit measures in predicting future choices of
in this research, we used implicit and explicit mea- decided and undecided individuals (for a review,
sures jointly to investigate automatic associations as see Gawronski & Galdi, 2011). These studies dem-
an alternative route for the emergence of stereo- onstrate that well-structured automatic associations
type-induced underperformance in young children. about a specic object are present also in the cogni-
tive maps of those individuals who, on explicit
measures, report being undecided and do
Insights From Social-Cognitive Research
not endorse well-dened conscious beliefs yet.
Implicit measures, such as the well-known Impli- These ndings come from research on attitudes
cit Association Test (IAT; Greenwald, McGhee, & (i.e., dealing with the affective evaluation of an
Schwartz, 1998), have been developed to assess the object; Eagly & Chaiken, 1993), whereas stereotypes
strength of automatic associations between concepts are cognitive beliefs about the link between a cate-
or concepts and attributes. These procedures are gory and an attribute (Hamilton & Trolier, 1986);
based on the rationale that individuals should be nonetheless they suggest that in terms of emer-
faster at pairing concepts, or concepts and attributes gence, automatic associations may precede conscious
that are strongly associated in their cognitive map, beliefs. Thus, there is reason to believe that implicit
as compared to those that are weakly or not at all measures can also detect recently formed automatic
associated. Research has shown that men and associations between concepts and stereotypical
women differ in the strength of their automatic attributes (e.g., boysmath) that are not reected on
associations between gender and academic domains explicit measures yet.
(Nosek, Banaji, & Greenwald, 2002) and that auto-
matic associations consistent with the mathgender
Insights From Research on Malleability of Automatic
stereotype predict attitudes toward math, domain
Associations
identication, and math performance (e.g., Kiefer &
Sekaquaptewa, 2007; Nosek & Smyth, 2011). Other research has demonstrated that external
Together, these results show that automatic associa- cues (e.g., situational stimuli) may increase or
tions play an important role in peoples academic decrease the activation of automatic associations in
achievement, interest, and performance, and that adults. For example, Dasgupta and Asgari (2004)
the combined use of implicit and explicit measures found that participants exposed to counterstereo-
may allow researchers to provide useful insights typical women showed lower activation of auto-
that neither implicit measures nor explicit measures matic associations consistent with genderrole
alone would offer. stereotypes, as compared to control participants.
Theorizing on the meaning of implicit and Similarly, women exposed to gender-stereotypical
explicit measures posits that explicit measures women in TV commercials showed increased acti-
reect more recent as well as newly formed vation of automatic associations consistent with
representations that are not strong enough yet to the traditional female stereotype, and this higher
Stereotype Threat Before Stereotype Awareness 253

activation of automatic associations led to worse To our knowledge, only Ambady and collabora-
math performance (Davies, Spencer, Quinn, & Ger- tors used an implicit stereotype awareness task
hardstein, 2002). The last results are consistent with (Ambady et al., 2001, p. 387) to assess the math
research demonstrating that stereotypical automatic gender stereotype in a study on stereotype threat in
associations affect working memory resources young children. The measure was a memory recall
needed to perform complex cognitive tasks. Speci- task, in which the experimenter told a story about a
cally, Forbes and Schmader (2010) found that student especially good at math, and recorded
women trained via an IAT to associate their gender whether the participant, when repeating the story,
with being good at math showed higher working used the pronoun he or she to identify the protago-
memory and increased math performance. Taken nist. However, rather than a measure of stereotypi-
together, these ndings suggest that it may be spe- cal automatic associations, this task should be
cically in automatic associations between group considered as an indirect measure of childrens
membership and stereotypical attributes, and in the stereotypes. Therefore, it is still unknown whether
malleability of such automatic associations, where automatic associations may be responsible for ste-
we should look not only to investigate stereotype- reotype threat performance decits in young chil-
induced underperformance but also to reduce dren.
stereotype threat effects. Unfortunately, to date no In this study, we focused on rst-grade children
study has tested whether automatic associations are for two reasons. So far, the only study showing the
malleable in children as well, or whether changes presence of automatic associations consistent with
in activation of stereotypical automatic associations gender stereotypes about math and language in
may reect changes in performance. Therefore, we young children (Cvencek et al., 2011) considered
hypothesized that situational cues making a nega- children from 6 to 7 years altogether (Grades 1 and
tive stereotype salient may automatically activate 2). Second, studies showing stereotype-induced
corresponding automatic associations. The activa- math performance decrements in young girls have
tion of such stereotypical automatic associations addressed 5- to 7-year-old children (Ambady et al.,
should in turn affect the performance of children 2001; Tomasetto et al., 2011) or 7-year-olds (Neuville
who are stereotype targets, even in the absence of & Croizet, 2007). Thus, the focus on 6-year-olds
evidence for stereotype endorsement. Conversely, (Grade 1) allowed us to investigate stereotypical
cues challenging the negative stereotype should automatic associations as well as stereotype threat
reduce the activation of corresponding automatic effects specically when children enter elementary
associations, and such a reduction should lead to school, thus becoming acquainted for the rst time
improved performance. with formal teaching of math (at least in the Italian
school system).
The study had four goals. First, using a child-
Aims of This Study
friendly version of the IAT (ChildIAT; Baron &
To test the hypotheses mentioned above, we Banaji, 2006) we tested whether gender stereotypes
employed explicit and implicit measures jointly to about math and language are present as automatic
assess two aspects of the mathgender stereotype, associations in 6-year-olds. Consistent with Cvencek
stereotype endorsement and automatic associations, et al. (2011), we expected stereotype-consistent auto-
and their differential relation with young childrens matic associations for both girls and boys.
stereotype threat vulnerability. Second, we investigated the malleability of chil-
Research with children has already used implicit drens automatic associations. It was hypothesized
measures to assess automatic associations (e.g., that childrens stereotypical automatic associations
Baron & Banaji, 2006). Regarding mathgender ste- would increase or decrease in activation depending
reotypes, Steffens, Jelenec, and Noack (2010) found on the stereotype content of a prior manipulation
stereotypical automatic associations in 9- to 14-year- task. We predicted an increased activation of stereo-
old girls and no automatic associations for boys at typical automatic associations in a stereotype-consistent
any age. Conversely, Cvencek et al. (2011) found condition than in a stereotype-inconsistent condition,
stereotype-consistent automatic associations in both respectively, conrming or contradicting the tradi-
girls and boys as young as 67 years of age. How- tional gender stereotype about math, with results in
ever, neither Steffens et al. nor Cvencek et al. the middle in a control condition, in which the math
assessed either performance or automatic associa- gender stereotype was not salient.
tions under stereotype threat. Therefore, these stud- Third, using an explicit measure, we investigated
ies are not directly relevant to our main hypothesis. whether 6-year-olds endorse the belief that boys are
254 Galdi, Cadinu, and Tomasetto

better than girls at math. Three reasons led us to to describe the picture, and then left the picture on
assess stereotype endorsement: (a) children rst one side of the desk. After the manipulation task,
develop stereotype endorsement and then stereo- children were invited to play a computer game (i.e.,
type awareness (e.g., Aboud, 1988; Martin & Ruble, the ChildIAT) using a laptop computer. Partici-
2010), (b) 6-year-olds do not possess stereotype pants were told that they would see pictures during
awareness yet (e.g., Rholes & Ruble, 1984; Selman, the game, and that they would press the red (A key)
1980), and (c) childrens personal stereotypic beliefs or the green (L key) button of the computer board to
could be sufcient to trigger stereotype threat, spe- indicate which picture they saw. After the Child
cically because at this age children assume that IAT, children performed eight math calculations
their beliefs are shared by others as well (e.g., (i.e., math test). The experimenter asked one calcula-
Augoustinos & Rosewarne, 2001). Consistent with tion at a time (e.g., How much is 5 plus 5?) and
previous research (e.g., Ambady et al., 2001; Muzz- children had to respond within 5 s. For each ques-
atti & Agnoli, 2007), we expected no evidence of tion, the experimenter noted whether the response
endorsement of the mathgender stereotype. was correct or not (i.e., incorrect or no response by
Finally, like in past research on stereotype threat 5 s), without providing any feedback. Finally, chil-
including 6-year-olds (e.g., Tomasetto et al., 2011), dren performed the explicit stereotype-endorsement
we expected girls math performance to decrease task. They were shown a picture of a boy and a girl
under stereotype threat, that is, in the stereotype- side by side and were told: These are a boy and a
consistent as compared to the control condition and girl. They are 6-year-olds and they are good at
the stereotype-inconsistent condition, which should school. Is the boy better at math, the girl better at
lead to the highest performance. Importantly, we math, or are they the same at math? Next, the
predicted the increased activation of stereotype- experimenter recorded the response, and partici-
consistent automatic associations to mediate the pants were thanked, given a candy bar, and dis-
decrease in girls performance under stereotype missed.
threat. Thus, the administration of the implicit measure
always followed the experimental manipulation and
always preceded the math test followed by the expli-
Method cit measure. This order of the tasks was chosen
because we aimed at testing the effects of the experi-
Participants
mental manipulation on automatic associations
Two hundred and seventy-six rst-grade chil- ruling out the possibility that counterbalancing
dren (143 girls) participated in the study. All of implicit and explicit measures might weaken the
them were born in Italy and were of typical age for effects of the experimental manipulation on auto-
their grade (M = 77.60 months, SD = 0.28). Chil- matic associations. Moreover, there is no evidence
dren attended one of six elementary schools in one that performing the IAT before a self-report induces
of three small towns in the northeast of Italy. Per- reactance or assimilation tendencies in the subse-
missions to conduct the study were granted by quent self-report (Nosek, Greenwald, & Banaji, 2005).
school principals and parents. Six female experi-
menters were involved in the data collection, con-
Materials
ducted during school hours. All children were
tested in the same school period to avoid potential Experimental manipulation. According to the
effects of variability in mathematics curricula across stereotype threat model, it is the salience of a nega-
classes at the moment of data collection. tive domain-relevant in-group stereotype that
impairs targets performance. A negative domain-
relevant in-group stereotype can be made salient in
Procedure
three ways: (a) describing the task that targets will
Children were tested individually in a quiet room subsequently perform as diagnostic of the ability
of their school, while sitting at a desk facing the related to the negative in-group stereotype, (b) mak-
experimenter. Each experimental session started by ing targets in-group salient, or (c) making the con-
asking participants to color one of three pictures, tent of the negative stereotype salient (Inzlicht &
depending on the experimental condition (stereo- Schmader, 2012). In this study, the manipulation
type consistent, control, and stereotype inconsistent) choice was to make salient the stereotype content
to which they were randomly assigned. When chil- (mathgender stereotype). Thus, participants were
dren nished coloring, the experimenter asked them randomly assigned to one of three experimental
Stereotype Threat Before Stereotype Awareness 255

conditions by inviting them to color a picture, in gory randomly presented twice) appeared on the
which: (a) a boy correctly resolves a math calculation screen. In this case, children pressed the red key
on a blackboard, whereas a girl fails to respond when they saw either a girl or a reading- or writing-
(stereotype-consistent condition); or (b) a girl related object, and the green key when they saw
correctly resolves the calculation, whereas a boy fails either a boy or a math-related object. In the fourth
to respond (stereotype-inconsistent condition); or block, participants categorized again the pictures
(c) a landscape was depicted (control condition). representing the categories mathematics and lan-
Implicit measure: The math(language)-gender Child guage. However, different from the key assignment
IAT. A ChildIAT (Baron & Banaji, 2006) was used in the second block, participants had to press the
to assess the relative strength of automatic associa- red key when they saw a math-related object and
tions between the target category boy and the the green key when they saw a reading- or writing-
attribute category mathematics, and the target cate- related object. Finally, in the fth double-categoriza-
gory girl and the attribute category language, as tion block, children categorized the same pictures of
compared to the opposite pairings (i.e., boylan- boys and girls and the same pictures of math-related
guage and girlmathematics). and reading- or writing-related objects. However,
Categories and stimuli for the ChildIAT. Four pic- participants now pressed the red key when they saw
tures of math-related objects (i.e., numbers) and four either a girl or a math-related object, and the green
pictures of reading- and writing-related objects (i.e., key when they saw either a boy or a reading- or
letters), pretested with rst-grade children for famil- writing-related object. Following the procedure
iarity and comprehension, were used as stimuli for employed in the third block, each picture represent-
the attribute categories mathematics and language. ing each of the four categories was randomly pre-
Four face-pictures of boys and four face-pictures of sented twice. The order of the two critical blocks,
girls represented the target categories boy and girl. third and fth, was counterbalanced across partici-
Procedure of the ChildIAT. Children were instruc- pants to avoid order effects.
ted to respond fast and accurately to three simple- Math test. Three days before the data collection,
categorization (practice) blocks and two (third and a math test was developed in collaboration with all
fth) critical double-categorization blocks. Each teachers of the classes participating in the study, to
practice block included 16 trials; each critical block create a set of difcult math calculations taking into
included 32 trials. On each trial, a target or attribute account the mathematics achievement of all classes.
picture appeared on the screen until participants The math test was a retrieval of numerical facts
gave their response. Children categorized each pic- including ve additions and three subtractions (i.e.,
ture by pressing a key on the computer board, the 5 + 5, 8 4, 6 + 4, 10 5, 2 + 3, 2 + 2, 6 3, and
left-hand response A or the right-hand response L 4 + 5). Children had to resolve each math calcula-
key. To simplify the task, the left-hand key (A) was tion in 5 s.
colored in red, and the right-hand key (L) was col- Explicit stereotype endorsement. As in Ambady
ored in green. The intertrial interval was 200 ms. A et al.s (2001) study, children were presented with a
red cross, which remained on the screen for 200 ms, photograph of a boy and a girl described as good
followed incorrect responses. at school, and had to choose whether the boy or the
In the rst block of the task, children were pre- girl is especially good at math or whether they are
sented with pictures for the target categories girl equal at math. The two photographs as well as the
and boy, with each picture randomly presented questions wording for the boy or the girl were
twice. Participants had to indicate whether the pic- presented in counterbalanced order.
ture on the screen was a boy or a girl by pressing
the red key for girl and the green key for boy. In
the second block, children were presented with pic- Results
tures for the attribute categories language and
Preliminary Analyses
mathematics, with each picture randomly presented
twice. Participants had to press the red key when Twenty-three girls and thirteen boys were
the picture was a reading- or writing-related object excluded only because of 30% or higher error rates in
and the green key when the picture was a math- at least one critical (double-categorization) block of
related object. In the third double-categorization the ChildIAT. Preliminary analyses conrmed that
block, both pictures of boys and girls, and pictures excluded participants were evenly distributed across
of math-related and reading- or writing-related conditions (i.e., 13 in the stereotype-consistent, 10 in
objects (with each picture representing each cate- the control, and 13 in the stereotype-inconsistent
256 Galdi, Cadinu, and Tomasetto

conditions), and that their responses did not differ Table 1


from those of retained children on both the math test Zero-Order Correlations Between Automatic Associations, Math Per-
and the stereotype endorsement. The nal sample formance, and Explicit Stereotype Endorsement for Girls and Boys
included 240 children (120 girls). 1 2 3
Individual scores of automatic associations were
calculated by means of the D-algorithm, designed for Girls (n = 120)
analyzing data with the IAT (Greenwald, Nosek, & Automatic associations
Banaji, 2003). The D-algorithm compares the extent Math performance .267**
to which performance on the incompatible critical Explicit stereotype endorsement .115 .142
Boys (n = 120)
block (i.e., girlmathematics and boylanguage shar-
Automatic associations
ing the same response key) is impaired relative to
Math performance .045
performance on the compatible critical block (i.e., Explicit stereotype endorsement .000 .034
girllanguage and boymathematics sharing the
same response key), taking into account participants Note. Pearsons r coefcients are reported for automatic associations
individual response latencies, standard deviations of and math performance. Spearmans rs coefcients are reported
for the categorical variable explicit stereotype endorsement.
latencies, and error rates in each of the two blocks. **p < .01.
Scores were calculated so that positive scores reect
stronger boymathematics and girllanguage automatic
associations. To estimate reliability of the ChildIAT, Zero-order correlations between all dependent
two ChildIAT scores were calculated: one using the variables (automatic associations, math performance,
rst half and the other using the second half of the and explicit mathgender stereotype endorsement),
two critical blocks. Internal consistency was satisfac- separately for girls and boys across experimental
tory (Cronbachs a = .84). conditions, are presented in Table 1.
A one-way analysis of variance (ANOVA) was
then conducted on scores of automatic associations
Automatic Associations and Their Malleability
using conditions (stereotype consistent, control, ste-
reotype inconsistent), gender (male, female), and We predicted stereotype-consistent automatic
the order of administration of the critical blocks associations for both girls and boys in the control
(1 = third block: girllanguage and boymath, fth block: condition, in which the mathgender stereotype
girlmath and boylanguage; 2 = third block: girlmath was not made salient. In addition, we aimed at test-
and boylanguage, fth block: girllanguage and boy ing for the malleability of childrens automatic asso-
math) as the independent variables. A main effect for ciations. Higher scores of automatic associations
order of administration was found, F(1, 234) = 6.64, were expected in the stereotype-consistent than in
p = .01, gp2 = .03. However, because order of the stereotype-inconsistent condition, with results in
administration produced no signicant interactions the middle for the control condition.
(all ps > .09), it is not further discussed. A two-way ANOVA was conducted on scores of
For each participant, correct responses to the automatic associations, with condition (stereotype
math test (i.e., correct math calculation in 5 s) were consistent, control, stereotype inconsistent) and
coded +1, and incorrect responses (i.e., no response gender (male, female) as the between-participants
or incorrect math calculation in 5 s) were coded 0. variables. Results showed a signicant Condition 9
For each child, responses were then added up in a Gender interaction, F(2, 234) = 6.35, p = .002,
single math score. To correct for potential variabil- gp2 = .05. To assess the effect of condition within
ity in mathematics prociency across classes, math gender, simple-effect analyses were conducted sepa-
scores were standardized within each class of the rately for girls and boys. Table 2 presents the aver-
six schools. Internal consistency of the math test age scores of automatic associations in each
based on all eight calculations was satisfactory experimental condition for both boys and girls,
(Cronbachs a = .89). together with information about the planned con-
Separate one-way ANOVAs were conducted on trasts between the means.
scores of automatic associations and math scores, No effect of condition emerged for boys
using condition (stereotype consistent, control, ste- automatic associations (p > .30). Moreover, one-sam-
reotype inconsistent), gender (male, female), and ple t tests against 0 showed no effect in any condi-
the schools as independent variables. Schools did tion (all ps > .20). Conversely, simple-effect analysis
not lead to any signicant result (all ps > .70), and on girls automatic associations revealed a signi-
thus this variable is not further discussed. cant effect of condition, F(2, 117) = 7.75, p < .01,
Stereotype Threat Before Stereotype Awareness 257

Table 2 the stereotype-consistent and the control conditions


Average Scores of Automatic Associations and Math Scores as a Func- than the stereotype-inconsistent conditions. No effect
tion of Condition (Stereotype Consistent, Control, Stereotype Inconsis- of the experimental manipulation was found for
tent) for Girls and Boys
boys automatic associations.
Stereotype Stereotype
consistent Control inconsistent
Math Performance
Girls (n = 120) For girls, we expected performance decrements
Automatic **0.28a *0.19a 0.10b in the stereotype-consistent condition as compared
associations (SD = 0.45) (SD = 0.78) (SD = 0.43)
to the control and the stereotype-inconsistent condi-
Math test 0.36a 0.01ab 0.16b
tions. Conversely, no effect of condition on boys
(SD = 0.73) (SD = 0.49) (SD = 0.92)
Boys (n = 120)
math performance was hypothesized.
Automatic 0.03a 0.05a 0.10a An ANOVA on math scores, with condition
associations (SD = 0.44) (SD = 0.30) (SD = 0.50) (stereotype consistent, control, stereotype inconsis-
Math test 0.28a 0.05a 0.03a tent) and gender (male, female) as the between-
(SD = 1.03) (SD = 1.02) (SD = .75) participants variables, was conducted. A Condi-
tion 9 Gender interaction emerged, F(2, 234) = 4.69,
Note. Means within rows not sharing the same subscript are sig- p = .010, gp2 = .04. Therefore, the effect of condition
nicantly different from each other at the p < .05 level (Bonfer-
roni test). was tested separately for girls and boys. Simple-
*p < .02. **p < .001. effect analyses on boys math scores (Table 2)
showed that boys performed equally well across
gp2 = .06, linear trend, p < .001. Post hoc tests with conditions (p > .30). To the opposite, an effect of
Bonferroni correction of the alpha level for multiple condition was found for girls, F(2, 117) = 3.66,
comparisons showed that girls scores of automatic p < .03, gp2 = .03, linear trend, p < .01: Girls under-
associations were lower in the stereotype-inconsis- performed in the stereotype-consistent as compared
tent as compared to the control (p < .02) and the to the stereotype-inconsistent condition (p < .03). On
stereotype-consistent (p < .01) conditions. As shown the contrary, math scores in the control and both in
in Table 2, although girls automatic associations the stereotype-consistent and in the stereotype-
were the highest in the stereotype-consistent inconsistent conditions were not different from each
condition, scores in the control and the stereotype- other (both ps > .20). Thus, whereas the salience of
consistent conditions did not differ from each other the negative in-group stereotype led girls to perform
(p > .90). Moreover, one-sample t tests against 0 worse on the math test, the removal of the same
showed that girls scores of automatic associations stereotype led girls to perform better in the
were different from 0 in the stereotype-consistent stereotype-inconsistent condition as compared to
and in the control (both ps < .02) but not in the the stereotype-consistent condition.
stereotype-inconsistent condition (p = .20).
Importantly, simple-effect analyses on the main
Explicit Stereotype Endorsement
effect of gender within conditions revealed that
boys scores were lower than girls scores of To assess stereotype endorsement, participants
automatic associations in the stereotype-consistent, were shown a picture of a boy and a girl and were
F(1, 234) = 5.90, p < .03, gp2 = .02, and control con- asked to say whether the boy or the girl is espe-
dition, F(1, 234) = 5.84, p < .03, gp2 = .02. Although cially good at math, or whether they are equal. For
this difference fell short of signicance, boys scores each child, stereotype-inconsistent response (i.e., the
of automatic associations tended to be higher than girl is better at math) was coded 1, neutral choice
those of girls in the stereotype-inconsistent condi- (i.e., the boy and the girl are equal) was coded 0, and
tion, F(1, 234) = 3.79, p = .053, gp2 = .01. stereotype-consistent choice (i.e., the boy is better at
In sum, although automatic associations consis- math) was coded +1. We expected no evidence that
tent with math(language)gender stereotypes were children endorsed the mathgender stereotype.
not found for boys, girls revealed stereotype-consis- A logistic regression for ordinal dependent mea-
tent automatic associations in both the control and sures was performed on childrens choices at the task
the stereotype-consistent conditions. Importantly, ( 1 = stereotype-inconsistent, 0 = neutral, +1 = stereo-
girls automatic associations showed the expected type-consistent choice), with gender (0 = male,
malleability, thus differing in activation as a function 1 = female), condition ( 1 = stereotype inconsistent,
of the experimental condition: They were higher in 0 = control, +1 = stereotype consistent), and the product
258 Galdi, Cadinu, and Tomasetto

term as predictors. A main effect of gender emerged,


Wald v2 = 11.49, p < .001, indicating that the proba- Automatic Associations

bility of a counterstereotypical versus neutral versus -.36**


stereotypical response was not equal for boys and -.37***

girls. Regardless of condition, 57% of boys and 57%


of girls favored their own gender and indicated the
Experimental .38* (.55**)
boy or the girl, respectively, as the most talented for Condition
Math Performance

math. Only 12% of boys and 21% of girls responded


that the boy and the girl were the same. Of the
Figure 1. Results of mediation analyses testing indirect effects of
remaining children, 22% of girls identied the boy experimental condition (stereotype consistent = 0, stereotype incon-
as being better at math, whereas 31% of boys sistent = 1) on math performance via automatic associations in
believed that the outstanding math student was the girls (n = 120).
girl. Thus, consistent with predictions, 6-year-olds *p < .05. **p < .01. ***p < .001.
did not endorse the mathgender stereotype, but
simply manifested gender in-group favoritism.
summarizes results for Contrast 1. To test for media-
tion, we calculated a bias-corrected 95% condence
Mediation Analysis
interval for the indirect effect (.13, SE = .06) using a
Because of null ndings for boys on both auto- bootstrapping technique (Preacher & Hayes, 2008).
matic associations and math scores, a mediation As the null hypothesis of no mediation states that
analysis was conducted only for girls to test the indirect effect is 0, the null hypothesis is rejected
whether automatic associations mediated the rela- when the condence interval does not include 0. In
tion between condition and math performance. the present analysis, the condence interval (with
Given that the predictor (i.e., condition) was a cate- 5,000 resamples) for the estimate of the indirect
gorical variable with three levels, we created two effect of Contrast 1 on performance did not include
dummy-coded variables to conduct the mediation 0, 95% CI [0.02, 0.28], thus conrming that auto-
analysis (Hayes & Preacher, 2012) with the stereo- matic associations mediated the relation between
type-consistent condition as the reference group. condition (stereotype consistent vs. stereotype
Specically, Contrast 1 tested the effect of the ste- inconsistent) and girls performance.
reotype-consistent (coded 0) versus stereotype-
inconsistent (coded 1) condition, with the control
Supplementary Analyses
condition coded 0. Contrast 2 tested for the residual
difference between the stereotype-consistent (coded Although main analyses showed that stereotype
0) and the control (coded 1) conditions, with the endorsement did not vary as a function of
stereotype-inconsistent condition coded 0. condition, supplementary analyses were conducted
Consistent with the univariate analyses reported to test whether the relations among condition,
above, the effect of Contrast 1 on performance was girls automatic associations, and girls math per-
signicant, b = .55, t(111) = 3.38, p < .01, whereas formance were further moderated by stereotype
the effect of Contrast 2 fell short of signicance, endorsement. Two-way analyses of variance were
b = .34, t(111) = 1.91, p = .06. Similarly, the effect of conducted on both scores of automatic associations
Contrast 1 on automatic associations was signicant, and math performance, with condition (stereotype
b = .37, t(111) = 3.56, p < .001, whereas the effect consistent, control, stereotype inconsistent) and
of Contrast 2 was not, p > .40. Moreover, when explicit stereotype endorsement (stereotype-consis-
automatic associations and the two contrasts were tent, neutral, stereotype-inconsistent response) as
entered simultaneously in the model predicting the between-participants variables. In both
performance, the effect of automatic associations ANOVAs, explicit stereotype endorsement yielded
was signicant, b = .36, t(111) = 2.32, p < .03, indi- no signicant effects: The effect of condition was
cating that automatic associations negatively affect signicant on both scores of automatic associa-
performance. Importantly, the effect of Contrast 1 tions, F(2, 111) = 7.76.12, p < .001, and math per-
(i.e., stereotype-consistent vs. stereotype-inconsistent formance, F(2, 111) = 4.12, p < .02, and was not
condition) was reduced, b = .38, t(110) = 2.06, further qualied by explicit stereotype endorse-
p < .05, whereas the effect of Contrast 2 (i.e., stereo- ment (all ps > .15). However, given that the rela-
type-consistent vs. control condition) remained not tively low number of girls who provided a neutral
signicant, b = .31, t(111) = 1.77, p = .07. Figure 1 response or indicated the other gender group as
Stereotype Threat Before Stereotype Awareness 259

better at math resulted in unbalanced cell frequen- the emergence of stereotype awareness, and in the
cies, this nding could be a consequence of low absence of endorsement of the mathgender stereo-
statistical power. To rule out this possibility, we type (Ambady et al., 2001; Neuville & Croizet,
repeated the same analyses by recoding explicit 2007; Tomasetto et al., 2011). The present research
stereotype endorsement into a dichotomous vari- disentangles this paradox and extends the stereo-
able, comparing girls indicating their gender group type threat model by demonstrating that automatic
as better at math with those who either indicated associations consistent with a negative in-group ste-
boys as better or provided a neutral response. reotype represent a key factor that may trigger
Results were very similar: The effect of condition stereotype threat in young children.
remained signicant on both automatic associa- Consistent with research showing that automatic
tions, F(2, 114) = 6.83, p < .01, and math perfor- associations can precede conscious beliefs (e.g., Gaw-
mance, F(2, 114) = 4.59, p < .01, and was not ronski & Galdi, 2011), we demonstrated for the rst
further qualied by explicit stereotype endorse- time that girls acquire the mathgender stereotype as
ment (all ps > .09). automatic associations before its emergence at the
We also tested whether automatic associations may explicit level: Despite the absence of evidence that
act as a moderator, rather than a mediator, of the children endorsed the mathgender stereotype,
effect of stereotype activation on girls math perfor- 6-year-old girls, but not boys, even in the control
mance. An analysis of covariance (ANCOVA) was condition (i.e., in the absence of any experimental
performed on math performance with condition manipulation aimed at increasing or decreasing the
(stereotype consistent, control, and stereotype incon- salience of the stereotype) showed stereotype-consis-
sistent), automatic associations, and the interaction tent automatic associations between boy and mathe-
between condition and automatic associations as matics, and girl and language. Moreover, consistent
predictors. The main effect of automatic associations with research showing malle-ability of automatic
was signicant, F(1, 114) = 5.57, p < .02, whereas the associations in adults (e.g., Dasgupta & Asgari,
interaction between condition and automatic associa- 2004), for the rst time, malleability of girls auto-
tions was not (p > .20), thus conrming that auto- matic associations was found. After coloring a draw-
matic associations acted as a mediator of the relation ing of a boy correctly solving a math problem, girls
between condition and performance. stereotypical automatic associations were activated,
Finally, we tested whether automatic associations as compared to the control condition. At the same
and stereotype endorsement may interact with each time, after coloring a picture of a girl succeeding in
other in determining girls vulnerability to stereo- math, stereotypical automatic associations were
type threat. We carried out a three-way ANCOVA reduced.
on math performance with condition (stereotype Importantly, the activation of automatic associa-
consistent, control, stereotype inconsistent), auto- tions in the stereotype-consistent condition ham-
matic associations, and explicit stereotype endorse- pered girls math performance as compared to the
ment response (stereotype consistent, neutral, stereotype-inconsistent condition, in which, con-
stereotype inconsistent) as predictors. Again, only versely, the reduced activation of automatic associa-
the main effect of automatic associations was signif- tions led girls to the highest performance.
icant, F(1, 114) = 4.01, p < .05, whereas no other Furthermore, highlighting the role of counterstereo-
main effect or higher order interaction attained sig- typical information in counteracting stereotype
nicance (all ps > .10). threat effects, these ndings are parallel to research
Although boys showed null ndings on all showing that providing information about equal
dependent variables, the same supplementary anal- gender abilities in math (i.e., stereotype-inconsistent
yses were conducted for them as well. In all analy- information) is an effective strategy to prevent
ses, neither main effects nor interactions reached womens performance decits (e.g., Cadinu, Maass,
signicance (all ps > .08). Frigerio, Impagliazzo, & Latinotti, 2003).
An open issue of this study concerns the poten-
tial processes underlying girls underperformance in
the stereotype-consistent versus -inconsistent condi-
Discussion
tion. Consistent with Forbes and Schmader (2010),
Although the stereotype threat model identies ste- one may speculate that activated stereotypical auto-
reotype awareness as a requirement for stereotype matic associations in the stereotype consistent con-
threat effects, research has shown that gender iden- dition may have burdened working memory capacity,
tity activation affects girls math performance before thus disrupting subsequent math performance.
260 Galdi, Cadinu, and Tomasetto

Conversely, the reduced activation in the stereo- example, ample evidence shows that children are
type-inconsistent condition may have freed up sensitive to adults nonverbal behavior and that such
working memory resources, thus enhancing subse- behaviors, often occurring in automatic and uncon-
quent performance. However, no such conclusions scious ways, may represent an important channel
can be drawn as this study did not test how girls through which attitudes are transmitted (e.g., Rud-
automatic associations affect performance. There- man, 2004; Walden & Ogan, 1988). Thus, even
fore, a direct test of the working memory hypothe- though adults may explicitly encourage nonstereo-
sis should be the goal of future studies. typical interests in children, they may exhibit differ-
Another potential process underlying girls worse ent patterns of nonverbal behaviors: Parents may
performance in the stereotype-consistent than the purchase more games or manipulative materials
stereotype-inconsistent condition could be a mere related to math and science for their sons than for
priming effect, which would lead individuals to their daughters (Jacobs & Bleeker, 2004), or intrude
behave automatically consistent with a cognitively more often in their daughters than in their sons
activated image (e.g., Wheeler & Petty, 2001). How- math homework to offer unsolicited help (e.g.,
ever, if a priming effect were at work, one should Bahnot & Jovanovic, 2005). As a result, such repeated
expect underperformance even for boys in response co-occurrences of objects and events might promote
to stereotype-related stimuli. To the contrary, nei- the automatic formation of stereotype-consistent
ther boys automatic associations nor performance associative links in childrens cognitive maps.
were affected by the activation of the mathgender In contrast to girls, boys did not reveal stereotypi-
(counter)stereotype. Thus, differently from a prim- cal automatic associations in any conditions.
ing effect, we argue that in this study it was the Consistent with research demonstrating that elemen-
membership in a negatively stereotyped group tary school boys tend to manifest in-group favoritism
(whose stereotype has been acquired via automatic regarding both math and language (Steele, 2003),
associations) that was responsible for girls perfor- these results could simply reect the fact that boys
mance decit under stereotype threat. showed in-groupserving automatic associations
Although girls automatic associations increased about both math and language domains (Steffens
and decreased in activation depending on the ste- et al., 2010), regardless of the stereotype content con-
reotype content of the experimental condition, no veyed by the experimental condition. Indeed, the
changes emerged at the level of endorsement of ste- presence of in-group-serving boymathematics and
reotypical beliefs. Consistent with research showing boylanguage automatic associations would result in
gender in-group bias in young children, the major- fast response latencies in both critical blocks of the
ity of children across conditions indicated their gen- ChildIAT. Thus, given that the D-algorithm com-
der as superior in math. We argue that this lack of pares the extent to which performance on the incom-
endorsement of the mathgender stereotype further patible (i.e., girlmath and boylanguage sharing the
strengthens the role of automatic associations in the same response key) is impaired relative to the
process of stereotype acquisition in children. compatible critical block (i.e., girllanguage and
Relative to the latest point, the present pattern of boymath sharing the same response key), in-group
ndings raises the tricky issue concerning the serving automatic associations could have led to the
sources and mechanisms underlying the formation small or null score of automatic associations that was
of automatic associations. Up to date, highly inuen- found for boys across conditions. However, because
tial theorizing has posited that automatic associa- both gender stereotypes regarding math and
tions stem from attitudes and conscious beliefs that language contribute to the ChildIAT score, and can-
have become overlearned and automatized over time not be separated (Nosek et al., 2005), this possibility
(e.g., Rudman, 2004; Wilson et al., 2000). However, could not be tested in this study. To our knowledge,
recent research (e.g., Gawronski & Galdi, 2011), as only one study used an implicit measure to assess
well as the present results, suggests that also other mathgender and languagegender stereotypes
underlying mechanisms may be responsible for the separately (i.e., Go-No-Go association task [GNAT];
formation of automatic associations. Drawing on the Nosek & Banaji, 2001) in a sample of 14-year-olds
distinction between associative and propositional and university students. Consistent with the in-
learning (Gawronski & Bodenhausen, 2006), we groupserving explanation, Steffens and Jelenecs
argue that the automatic formation of stereotype-con- (2011) male participants revealed both boysmath
sistent associative links in childrens cognitive map stereotypical automatic associations and boyslan-
could stem from repeated co-occurrences of objects guage counterstereotypical automatic associations.
or stimuli in childrens social environment. For However, as Steffens and Jelenec noted, the internal
Stereotype Threat Before Stereotype Awareness 261

consistency of the GNAT was low . . . and the IAT automatic associations as a novel prerequisite for
clearly appears more sensitive (Steffens & Jelenec, stereotype-induced underperformance in young
2011, p. 332). Thus, a goal of future studies should be children. Importantly, by showing also that chil-
to test whether other implicit measures that combine drens automatic associations are malleable, these
the measurement quality of the IAT with the separate ndings are promising in terms of interventions to
measurement of math and language gender stereo- promote gender equality in math and science
types, such as the GNAT, can be implemented and because they suggest that girls can be protected
adapted for use with young children. from the deleterious impact of mathgender stereo-
Studies on the development of cognitive compe- types. Relevant to this claim is a study (Dasgupta &
tencies offer alternative explanations for boys lack Greenwald, 2001) demonstrating that Caucasian
of automatic associations. For example, previous participants exposed to images of admired Black
research has shown earlier knowledge of gender and disliked White exemplars showed lower pro-
categories (Zosuls et al., 2009) and earlier achieve- White automatic associations than participants
ment of gender constancy (Ruble et al., 2007) by exposed to images of admired Whites and disliked
girls than boys. At the same time, differences in Blacks. Interestingly, such a decrease in pro-White
gender categorization abilities between girls and automatic associations lasted for 24 hr, suggesting
boys could also be the result of socialization pro- that relatively enduring changes in automatic associ-
cesses. For example, women show stronger auto- ations can be obtained. If so, repeatedly presenting
matic associations between the self, their gender, girls with exemplars of successful women in math
and the in-group gender stereotype as compared to and science could promote the reduction in stereo-
men (Cadinu & Galdi, 2012), and children from typical automatic associations, long before girls have
low-status groups are more likely to be aware of acquired any awareness or endorsement of the ste-
self-relevant stereotypes than children from high- reotype favoring males in math. This study suggests
status groups (McKown & Weinstein, 2003). that this strategy could protect girls performance in
Although consistent with Steffens et al. (2010), stereotype-threatening situations and potentially
the result that boys did not manifest stereotypical help them to expand their interests toward tradi-
automatic associations is inconsistent with Cvencek tionally male domains.
et al.s (2011) ndings, showing stereotypical auto-
matic associations between gender and academic
subjects in both 6- to 7-year-old boys and girls.
References
However, this discrepancy might reect cross-
cultural variations in the strength of gender stereo- Aboud, F. E. (1988). Children and prejudice. New York:
types at the societal level, which have been found Basil Blackwell.
to be associated with either smaller or bigger gen- Ambady, N., Shih, M., Kim, A., & Pittinsky, T. (2001).
der differences in the strength of stereotypical auto- Stereotype susceptibility in children: Effects of identity
activation on quantitative performance. Psychological
matic associations (Nosek et al., 2009), as well as in
Science, 12, 385390. doi:10.1111/1467-9280.00371
the endorsement of gender stereotypes about Aronson, J., & Good, C. (2003). The development and
academic abilities (e.g., Evans, Schweingruber, & consequences of stereotype vulnerability in adolescents.
Stevenson, 2002). Moreover, in a study involving In F. Pajares & T. Urdan (Eds.), Adolescence and educa-
Chinese, Japanese, and American children, Lummis tion: Vol. 2. Academic motivation of adolescents (pp. 299
and Stevenson (1990) found that whereas American 330). Greenwich, CT: Information Age.
and Chinese 6-year-olds do not believe that there Augoustinos, M., & Rosewarne, D. L. (2001). Stereotype
are gender differences in math, 6-year-old Japanese knowledge and prejudice in children. British Journal of
children tend to believe that boys are better than Developmental Psychology, 19, 143156. doi:10.1348/
girls at math (see also Del Rio & Strasser, 2013). 026151001165912
These results show that cultural differences in the Bahnot, R., & Jovanovic, J. (2005). Do parents academic
gender stereotypes inuence whether they intrude on
strength of gender stereotypes about academic abili-
their childrens homework? Sex Roles, 52, 597607.
ties, in schooling, and in out-of-school experiences, doi:10.1007/s11199-005-3728-4
may lead to different acquisition ages of the Baron, A. S., & Banaji, M. (2006). The development of
endorsement of these stereotypes, and suggest that implicit attitudes. Evidence of race evaluations from
the same could be true regarding the acquisition of ages 6 and 10 and adulthood. Psychological Science, 17,
stereotypical automatic associations. 5358. doi:10.1111/j.1467-9280.2005.01664.x
In conclusion, this study has the potential to con- Cadinu, M., & Galdi, S. (2012). The cognitive accessibility
tribute to the stereotype threat model by suggesting of womanhood: Implicit gender self-categorization
262 Galdi, Cadinu, and Tomasetto

leads to stronger self-stereotyping by women than men. Greenwald, A. G., McGhee, D. E., & Schwartz, J. K. L.
European Journal of Social Psychology, 42, 546551. doi: (1998). Measuring individual differences in implicit cog-
10.1002/ejsp.1881 nition: The implicit association test. Journal of Personality
Cadinu, M., Maass, A., Frigerio, S., Impagliazzo, L., & and Social Psychology, 74, 14641480. doi:10.1037/0022-
Latinotti, S. (2003). Stereotype threat: The effect of 3514.74.6.1464
expectancy on performance. European Journal of Social Greenwald, A. G., Nosek, B. A., & Banaji, M. R. (2003).
Psychology, 33, 267285. doi:10.1002/ejsp.145 Understanding and using the implicit association test: I.
Cvencek, D., Meltzoff, A., & Greenwald, A. G. (2011). An improved scoring algorithm. Journal of Personality
Math-gender stereotypes in elementary school children. and Social Psychology, 85, 187216. doi:10.1037/0022-
Child Development, 82, 766779. doi:10.1111/j.1467-8624. 3514.85.2.197
2010.01529.x Hamilton, D. L., & Trolier, T. K. (1986). Stereotypes and
Dasgupta, N., & Asgari, S. (2004). Seeing is believing: stereotyping: An overview of the cognitive approach. In
Exposure to counter-stereotypic women leaders and its J. Dovidio & S. Gaertner (Eds.), Prejudice, discrimination,
effect on the malleability of automatic gender stereo- and racism (pp. 127163). San Diego, CA: Academic
typing. Journal of Experimental Social Psychology, 40, 642 Press.
658. doi:10.1016/j.jesp.2004.02.003 Hayes, A. F., & Preacher, K. J. (2012). Statistical mediation
Dasgupta, N., & Greenwald, A. G. (2001). On the analysis with a multicategorical independent variable. Man-
malleability of automatic attitudes combating automatic uscript under review. Retrieved November 10, 2012,
prejudice with images of admired and disliked from http://www.afhayes.com/public/hp2012.pdf
individuals. Journal of Personality and Social Psychology, Inzlicht, M., & Schmader, T. (Eds.). (2012). Stereotype
81, 800814. doi:10.1037/0022-3514.81.5.800 threat: Theory, process, and application. New York: Oxford
Davies, P. G., Spencer, S. J., Quinn, D. M., & Gerhard- University Press.
stein, R. (2002). Consuming images: How television Jacobs, J. E., & Bleeker, M. M. (2004). Girls and boys
commercials that elicit stereotype threat can restrain developing interests in math and science: Do parents
women academically and professionally. Personality and matter? New Directions for Child and Adolescent Develop-
Social Psychology Bulletin, 28, 16151628. doi:10.1177/ ment, 106, 521. doi:10.1037/0022-0663.96.1.97
014616702237644 Katz, P. A., & Kofkin, J. A. (1997). Race, gender, and young
Del Rio, M. F., & Strasser, K. (2013). Preschool childrens children. In S. S. Luthar, J. A. Burack, D. Cicchetti, & J.
beliefs about gender differences in academic skills. Sex Weisz (Eds.), Developmental psychopathology: Perspectives
Roles, 68, 231238. doi:10.1007/s11199-012-0195-6 on adjustment, risk, and disorder (pp. 5174). New York:
Eagly, A. H., & Chaiken, S. (1993). The psychology of atti- Cambridge University Press.
tudes. Fort Worth, TX: Harcourt Brace Jovanovich. Kiefer, A. K., & Sekaquaptewa, D. (2007). Implicit stereo-
Evans, E. M., Schweingruber, H., & Stevenson, H. W. types, gender identication, and math-related out-
(2002). Gender differences in interest and knowledge comes: A perspective study of female college students.
acquisition: The United States, Taiwan, and Japan. Sex Psychological Science, 18, 1318. doi:10.1111/j.1467-9280.
Roles, 47, 153167. doi:10.1023/A:1021047122532 2007.01841.x
Forbes, C., & Schmader, T. (2010). Retraining attitudes Lummis, M., & Stevenson, H. W. (1990). Gender differ-
and stereotypes to affect motivation and cognitive ences in beliefs and achievement: A cross-cultural
capacity under stereotype threat. Journal of Personality study. Developmental Psychology, 26, 254263. doi:
and Social Psychology, 99, 740754. doi:10.1037/a0020971 10.1037/0012-1649.26.2.254
Gawronski, B., & Bodenhausen, G. V. (2006). Associative Martin, C., & Ruble, D. (2010). Patterns of gender devel-
and propositional processes in evaluation: An integra- opment. Annual Review of Psychology, 61, 353381.
tive review of implicit and explicit attitude change. Psy- doi:10.1146/annurev.psych.093008.100511
chological Bulletin, 132, 692731. doi:10.1037/0033-2909. Martinot, D., Bages, C., & Desert, M. (2012). French chil-
132.5.692 drens awareness of gender stereotypes about mathe-
Gawronski, B., & Galdi, S. (2011). Using implicit mea- matics and reading: When girls improve their
sures to read the minds of undecided voters. In M. reputation in math. Sex Roles, 66, 210219. doi:10.1007/
Cadinu, S. Galdi, & A. Maass (Eds.), Social perception, s11199-011-0032-3
cognition, and language in honour of Arcuri (pp. 203216). Martinot, D., & Desert, M. (2007). Awareness of a gender
Padova, Italy: CLEUP. stereotype, personal beliefs and self-perceptions regard-
Gawronski, B., & LeBel, E. P. (2008). Understanding pat- ing math ability: When boys do not surpass girls. Social
terns of attitude change: When implicit measures show Psychology of Education, 10, 455471. doi:10.1007/
change, but explicit measures do not. Journal of Experi- s11218-007-9028-9
mental Social Psychology, 44, 13551361. doi:10.1016/ McKown, C., & Strambler, M. J. (2009). Developmental
j.jesp.2008.04.005 antecedents and social and academic consequences of
Gawronski, B., & Payne, B. K. (Eds.). (2010). Handbook of stereotype-consciousness in middle childhood. Child
implicit social cognition: Measurement, theory, and applica- Development, 80, 16431659. doi:10.1111/j.1467-8624.
tions. New York: Guilford Press. 2009.01359.x
Stereotype Threat Before Stereotype Awareness 263

McKown, C., & Weinstein, R. (2003). The development Ruble, D., Taylor, L. J., Cyphers, L., Greulich, F. K., Lurye,
and consequences of stereotype consciousness in mid- L. E., & Shrout, P. (2007). The role of gender constancy
dle childhood. Child Development, 74, 498515. doi:10. in early gender development. Child Development, 78,
1111/1467-8624.7402012 11211136. doi:10.1111/j.1467-8624.2007.01056.x
Muzzatti, B., & Agnoli, F. (2007). Gender and mathemat- Rudman, L. A. (2004). Sources of implicit attitudes. Cur-
ics: Attitudes and stereotype threat susceptibility in rent Directions in Psychological Science, 13, 7982. doi:
Italian children. Developmental Psychology, 43, 747759. 10.1111/j.0963-7214.2004.00279.x
doi:10.1037/0012-1649.43.3.747 Selman, R. (1980). The growth of interpersonal understand-
Neuville, E., & Croizet, J. (2007). Can salience of gender ing: Developmental and clinical analysis. New York: Aca-
identity impair math performance among 78 years old demic Press.
girls? The moderating role of task difculty. European Steele, C. M. (1997). A threat in the air: How stereotypes
Journal of Psychology of Education, 22, 307316. doi: shape intellectual identity and performance. American
10.1007/BF03173428 Psychologist, 52, 613629. doi:10.1037/0003-066X.52.6.613
Nosek, B. A., & Banaji, M. R. (2001). The go/no-go associ- Steele, C. M., & Aronson, J. (1995). Stereotype threat and
ation task. Social Cognition, 19, 625666. doi:10.1521/ intellectual test performance of African Americans.
soco.19.6.625.20886 Journal of personality and Social Psychology, 69, 797811.
Nosek, B. A., Banaji, M. R., & Greenwald, A. G. (2002). doi:10.1037/0022-3514.69.5.797
Harvesting implicit group attitudes and beliefs from a Steele, J. (2003). Childrens gender stereotypes about
demonstration web site. Group Dynamics, 6, 101115. math: The role of stereotype stratication. Journal of
doi:10.1037/1089-2699.6.1.101 Applied Social Psychology, 33, 25872606. doi:10.1111/
Nosek, B. A., Greenwald, A. G., & Banaji, M. R. (2005). j.1559-1816.2003.tb02782.x
Understanding and using the Implicit Association Test: Steffens, M. C., & Jelenec, P. (2011). Separating implicit
II. Method, variables, and construct validity. Personality gender stereotypes regarding math and language:
and Social Psychology Bulletin, 31, 166180. doi:10.1177/ Implicit ability stereotypes are self-serving for boys and
0146167204271418 man, but not for girls and women. Sex Roles, 64, 324
Nosek, B. A., & Smyth, F. L. (2011). Implicit social cogni- 335. doi:10.1007/s11199-010-9924-x
tions predict sex differences in math engagement and Steffens, M. C., Jelenec, P., & Noack, P. (2010). On the
achievement. American Educational Research Journal, 48, leaky math pipeline: Comparing implicit math-gender
11251156. doi:10.3102/0002831211410683 stereotypes and math withdrawal in female and male
Nosek, B. A., Smyth, F., Sriram, N., Lindner, N., Devos, children and adolescents. Journal of Educational Psychol-
T., Ayala, A., et al. (2009). National differences in ogy, 102, 947963. doi:10.1037/a0019920
gender-science stereotypes predict national sex Thoman, D., White, P., Yamawaki, N., & Koishi, H.
differences in science and math achievement. Proceed- (2008). Variations of gender-math stereotype content
ings of the National Academy of Sciences of the United affect womens vulnerability to stereotype threat. Sex
States of America, 106, 1059310597. doi:10.1073/pnas. Roles, 58, 702712. doi:10.1007/s11199-008-9390-x
0809921106 Tomasetto, C., Alparone, F. R., & Cadinu, M. (2011).
Olson, M. A., & Fazio, R. H. (2006). Reducing automatically Girls math performance under stereotype threat: The
activated racial prejudice through implicit evaluative moderating role of mothers gender stereotypes. Devel-
conditioning. Personality and Social Psychology Bulletin, opmental Psychology, 47, 943949. doi:10.1037/a0024047
32, 421433. doi:10.1177/0146167205284004 Walden, T., & Ogan, T. (1988). The development of social
Perner, J., & Wimmer, H. (1985). John thinks that Mary referencing. Child Development, 59, 12301240.
thinks that: Attribution of second-order beliefs by 5- to Wheeler, S. C., & Petty, R. E. (2001). The effects of stereo-
10-year old children. Journal of Experimental Child Psychol- type activation on behavior: A review of possible mech-
ogy, 39, 437471. doi:10.1016/0022-0965(85)90051-7 anisms. Psychological Bulletin, 127, 797826. doi:10.1037/
Powlishta, K. K. (1995). Gender bias in childrens percep- 0033-2909.127.6.797
tions of personality traits. Sex Roles, 32, 1728. doi: Wilson, T. D., Lindsey, S., & Schooler, T. Y. (2000). A
10.1037/0012-1649.31.5.781 model of dual attitudes. Psychological Review, 107, 101
Preacher, K. J., & Hayes, A. F. (2008). Asymptotic and re- 126. doi:10.1037/0033-295X.107.1.101
sampling strategies for assessing and comparing indirect Yee, M., & Brown, R. (1994). The development of gender
effects in multiple mediator models. Behavior Research differentiation in young children. British Journal of Social
Methods, 40, 879891. doi:10.3758/BRM.40.3.879 Psychology, 33, 183196. doi:10.1111/j.2044-8309.1994.
Quintana, S. M. (1998). Childrens developmental under- tb01017.x
standing of ethnicity and race. Applied and Preventive Zosuls, K. M., Ruble, D., Tamis-LeMonda, C. S., Shrout,
Psychology, 7, 2745. doi:10.1016/S0962-1849(98)80020-6 P. E., Bornstein, M. H., & Greulich, F. K. (2009). The
Rholes, W. S., & Ruble, D. N. (1984). Childrens under- acquisition of gender labels in infancy: Implications for
standing of dispositional characteristics of others. Child sex-typed play. Developmental Psychology, 45, 688701.
Development, 55, 550560. doi:10.1037/a0014053