Cultural Diversity and Ethnic Minority Psychology © 2010 American Psychological Association

2010, Vol. 16, No. 1, 94 –98 1099-9809/10/$12.00 DOI: 10.1037/a0017564

Stereotype Boost and Stereotype Threat Effects:

The Moderating Role of Ethnic Identification

Brian E. Armenta
University of Nebraska—Lincoln

Belonging to a stereotyped social group can affect performance in stereotype-relevant situations, often
shifting performance in the direction of the stereotype. This effect occurs similarly for members of
positively and negatively stereotyped groups (i.e., stereotype boost and stereotype threat, respectively).
This study examined ethnic group identification as a moderator of these effects in the math performance
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

of Asian Americans and Latinos, who are positively and negatively stereotyped in this domain,
This document is copyrighted by the American Psychological Association or one of its allied publishers.

respectively. Results showed that high ethnically identified Asian Americans performed better and high
ethnically identified Latinos performed worse when an ethnicity– ethnic stereotype cue was present. The
performance of low ethnically identified Asian Americans and Latinos was not affected by this cue.

Keywords: stereotype threat, stereotype boost, ethnic identification, self-categorization

Simply being a member of a stereotyped group can affect activation of that identity and the associated stereotypes (e.g., Shih
performance on stereotype-relevant tasks (i.e., tasks for which the et al., 2002; Steele & Aronson, 1995). Thus, factors that increase
stereotype might apply). Specifically, when a stereotyped group the tendency that a stereotyped identity will become personally
identity and the associated group stereotypes are made salient, salient and cognitively activated should increase susceptibility to
performance tends to shift in the direction of the stereotype. This these effects.
occurs similarly for members of positively and negatively stereo- Self-categorization theory (Turner, Hogg, Oakes, Reicher, &
typed groups. For example, Shih, Pittinsky, and Ambady (1999) Wetherell, 1987) suggests that individuals who subjectively iden-
showed that Asian women performed better on a math exam when tify with their group are more likely to cognitively activate their
their ethnic identity was made salient (a stereotype boost effect; group identity in the presence of identity-relevant cues (Spears,
Shih, Ambady, Richeson, Fujita, & Gray, 2002) but performed Doosje, & Ellemers, 1999). This perspective suggests that the
worse on this exam when their gender identity was made salient stereotype boost and stereotype threat effects should be more
(a stereotype threat effect; Aronson & Steele, 1995), consistent pronounced among individuals who strongly identify with their
with the respective stereotypes about these groups. stereotyped group. In this study, I tested this prediction by exam-
The psychological mechanisms that account for these effects ining the math performance of Asian Americans, who are stereo-
have been and continue to be debated (e.g., Steele, Spencer, & typically viewed as mathematically talented (Niemann, Jennings,
Aronson, 2002; Wheeler & Petty, 2001). In general, however, Rozelle, & Baxter, 1994), and Latinos, who are stereotypically
stereotype threat is believed to result from increased concerns viewed as academically inept (Hunt & Espinoza, 2007), as a
about being evaluated in terms of a negative group stereotype function of ethnic identification and a situational cue that impli-
(Steele, 1997; Steele et al., 2002), whereas stereotype boost is cates their ethnicity and the associated ethnic group stereotypes
believed to result from an ideomotor process in which the mere regarding their group’s ability.
thought of an action, even if only at a nonconscious level, in- In contrast to self-categorization theory, it is possible that ethnic
creases the tendency to engage in that action (Dijksterhuis, Bargh, group identification will reduce or even reverse the stereotype
& Zanna, 2001; Wheeler & Petty, 2001). Regardless of the under- boost effect. For example, highly ethnically identified Asian
lying mechanisms, research has clearly demonstrated that the Americans may feel greater pressure to confirm the positive group
stereotype boost and stereotype threat effects occur when an en- stereotype as a means of maintaining a positive social identity
vironmental cue makes a group identity or the associated group (Tajfel & Turner, 1979). Ironically, this additional pressure may
stereotypes personally salient, ostensibly leading to the cognitive negatively affect their performance (Wheeler & Petty, 2001). Al-
ternatively, because the ethnic group stereotype is positive, high
and low ethnically identified Asian Americans may benefit equally
Brian E. Armenta, Department of Psychology, University of Nebraska. from situational cues that implicate their stereotyped identity. No
I thank Cynthia Willis Esqueda, Jennifer S. Gustavo Carlo, Carey S. studies have examined the role of group identification in the
Ryan, and Richard R. Dienstbier for their assistance with this article. I
stereotype boost effect; thus, this study provides an important
would also like to thank Lori Barker-Hackett, Jeffery S. Mio, and the
Psychology and Sociology Department at California State Polytechnic
examination of how group identification affects susceptibility to
University, Pomona, for providing the opportunity to conduct this study. this effect.
Correspondence concerning this article should be addressed to Brian E. There is preliminary evidence that group identification increases
Armenta, Department of Psychology, University of Nebraska, 238 Burnett susceptibility to the stereotype threat effect. Specifically, Schmader
Hall, Lincoln, NE 68588-0308. E-mail: [email protected] (2002) showed that women high in gender identification were more
94
 BRIEF REPORTS 95

vulnerable to the stereotype threat effect in mathematics. In contrast to White, 2001). This nine-item measure assesses the degree to which
these findings, and the prediction drawn from self-categorization math is important to one’s self-concept and consists of three types
theory, it is possible that ethnic identification functions differently of questions. Participants responded to four questions, such as
than gender identification in the stereotype threat effect. Indeed, a “How important is it to you to be good at mathematics?” on a
large body of research has demonstrated that ethnic identification can 5-point scale anchored by 1 (not at all) and 5 (very much).
buffer the negative effects associated with ethnic-based societal de- Participants also responded to four items, such as “Mathematics is
valuation and rejection (e.g., Armenta & Hunt, 2009; Armenta, one of my best subjects,” on a 5-point scale anchored by 1
Knight, Carlo, & Jacobson, 2008; Romero & Roberts, 2003; Umaña- (strongly disagree) and 5 (strongly agree). For the final item,
Taylor & Updegraff, 2007). Thus, ethnic identification may in fact participants responded to the question “Compared to other stu-
protect Latinos from the stereotype threat effect. No studies have dents, how good are you at math?” on a 5-point scale anchored by
examined the role of ethnic identification in the stereotype threat 1 (very poor) and 5 (excellent). These items formed a reliable scale
effect; thus, this study stands to make an important addition to the (␣ ⫽ .92).
literature. After completing the measures, participants were invited to
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

return to a second session. Those who agreed to return provided a

This document is copyrighted by the American Psychological Association or one of its allied publishers.

Method nondescriptive identification number and signed up for the second

session. Course credit was given separately for the completion of
each session.
Study Overview
This study was conducted in two sessions. During the first
session, Asian American and Latino undergraduate students com- Session 2: Experimental Procedure
pleted a measure of ethnic identification and math identification.
The second session took place in a medium-sized computer lab
Math identification has been shown to moderate the stereotype
(approximately 25 private computer stations), outside of class
boost and stereotype threat effects and was thus also assessed.
time. The racial composition of the testing sessions was not con-
During the second session, which took place 1–2 weeks later,
trolled or recorded. However, participants had little direct contact
participants completed a difficult math exam. Before taking this
with each other after arriving at the study. Specifically, a maxi-
exam, participants were randomly assigned to one of two
mum of 12 participants were allowed to complete the study at one
ethnicity– ethnic stereotype cue conditions: an experimental con-
session so that they could be seated at least one computer station
dition in which an ethnicity– ethnic stereotype cue was present and
apart from other participants. I provided verbal instructions to the
a control condition in which this cue was not present.
group as a whole. Participants were informed that they would
complete a math test and were given bogus information that their
Participants group had been assigned to take the paper version of the test. In
One hundred six Asian American (n ⫽ 66) and Latino (n ⫽ 40) fact, all tests in this study were completed in paper format. Par-
undergraduate students from a public Southern California univer- ticipants were not aware that they would be taking a math exam,
sity participated in this study for course credit. This sample con- specifically, until after they were seated and given verbal instruc-
sisted of 46 men and 60 women, with an average age of 20.5 years tions. The study was conducted in a computer lab to maintain an
(SD ⫽ 2.65). environment that was consistent with the purported study.
Participants were then given a packet with the math exam
enclosed. Participants were randomly assigned to receive one of
Session 1: Recruitment and Measure Completion two packets containing the experimental manipulation. For both
Participants were recruited from introductory psychology packets, information for the test was included (e.g., number of
courses for a study ostensibly examining the use of paper-based items, time allowed). In the experimental condition, participants
and computer-based testing methods. Participants read and signed read the following sentence: “[Understanding possible difference
an informed consent form, provided demographic information, and in testing methods] is crucial as it has been claimed that these types
completed several measures, including measures of ethnic identi- of tests measure individuals’ true intellectual ability, which his-
fication and math identification. Additional measures (e.g., self- torically have shown differences based on ethnic heritage.”
esteem) served as fillers. No mention was made at this point that In the control condition, this sentence was replaced with the
participants would be completing a math test during the second following: “[Understanding possible difference in testing methods]
session. is important because most standardized testing is moving toward
Ethnic identification was measured with the Affirmation and computer methods and away from the traditional paper and pencil
Belonging subscale of the Multi-Group Ethnic Identity Measure methods.”
(Phinney, 1992), which assesses the degree to which an individual After reading these instructions, participants were reminded that
feels positively attached to his or her ethnic group. Participants they would have 12 min to complete the test and that it was
responded to seven items, such as “I have a strong sense of important that they try to do their best. At this point, I started a
belonging to my ethnic group,” on a 5-point scale, anchored by 1 stopwatch and advised participants that they could begin. The math
(strongly disagree) and 5 (strongly agree). These items formed a test consisted of 14 questions on basic algebra and geometry,
reliable scale (␣ ⫽ .89). similar to questions found on the Graduate Record Exam. Each
Math identification was assessed with the Math Identification question had five response options, and participants were in-
subscale of the Domain Identification Measure (J. L. Smith & structed to circle the correct answer. A total math score was
 96 BRIEF REPORTS

computed by assigning 1 point to each correctly answered question Table 1

and summing across questions. Means (and Standard Deviations) by Ethnicity, Gender,
Finally, participants were questioned about suspicions regarding and Condition
the study. The most common response was that the study was
investigating the use of computers in math exams. No participants Ethnicity–ethnic stereotype cue
indicated suspicion about the true purpose of the study. After Asian Americans Latinos
questioning, participants were debriefed, thanked, and excused.
Gender No cue Cue No cue Cue
Results Male 5.79 (0.51) 7.13 (0.55) 6.80 (0.96) 5.25 (0.76)
Female 4.89 (0.51) 5.93 (0.56) 5.29 (0.58) 4.69 (0.60)
On average, participants attempted 11.44 of the 14 test items
(SD ⫽ 2.43) and correctly answered 5.65 of the items (SD ⫽ 2.23).
Interestingly, contrary to ethnic stereotypes (Hunt & Espinoza,
2007; Niemann et al., 1994), Asian Americans did not perform second step (stereotype boost–threat model). The two-way inter-
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

significantly better (M ⫽ 5.88, SD ⫽ 2.37) than Latinos (M ⫽

This document is copyrighted by the American Psychological Association or one of its allied publishers.

action between ethnicity and ethnicity– ethnic stereotype cue pro-

5.28, SD ⫽ 1.93), F(1, 104) ⫽ 1.89, p ⫽ .18. In addition, this vides a test of the stereotype boost and stereotype threat effects.
sample was moderately identified with mathematics (M ⫽ 3.26, The three-way interaction between ethnicity, ethnic identification,
SD ⫽ 0.93) and slightly more identified with their ethnic group and ethnicity– ethnic stereotype cue was entered on the final step
(M ⫽ 3.73, SD ⫽ 0.74). Also, contrary to ethnic stereotypes, Asian (moderation model). This interaction provides a test of the hypoth-
Americans and Latinos did not differ in their identification with esized moderation of the stereotype boost and stereotype threat
math (Ms ⫽ 3.31 and 3.19, SDs ⫽ 0.93 and .93, respectively), F(1, effects by ethnic identification. The results for these models are
104) ⫽ 0.45, p ⫽ .51. However, Latinos did identify slightly more shown in Table 2.
with their ethnic group (M ⫽ 3.99, SD ⫽ 0.70) than did their Asian
American counterparts (M ⫽ 3.58, SD ⫽ 0.73); F(1, 104) ⫽ 8.05, Step 1: Main Effects Model
p ⫽ .005. As would be expected, math identification was posi-
tively associated with math performance, r(104) ⫽ .24, p ⫽ .01; This model was significant (R2 ⫽ .12), F(5, 100) ⫽ 2.63, p ⫽
however, it did not interact with any other variables in predicting .03. As shown in Table 3, gender and math identification were the
math performance. Thus, I included math identification in the only significant predictors and followed the same pattern as re-
primary analyses as a control variable to isolate the effect of the ported earlier.
experimental manipulation that is not a result of previous math
identity (see Thoman, White, Yamawaki, & Koishi, 2008, for a Step 2: Stereotype Boost–Threat Model
similar use of domain identification). The inclusion of the two-way interactions resulted in a signifi-
Some research has shown that women are susceptible to the cant increase in explained variance (⌬R2 ⫽ .07), ⌬F(3, 97) ⫽ 2.65,
stereotype threat effect in math performance (e.g., Schmader, p ⫽ .05. The only significant interaction was between ethnicity
2002; Spencer, Steele, & Quinn, 1999). Thus, I considered gender and ethnicity– ethnic stereotype cue. I conducted follow-up tests
as an additional factor in this study. Results showed that men using planned contrast analyses. To test the expected stereotype
performed significantly better (M ⫽ 6.24) than women (M ⫽ 5.20) boost effect, a weight of 3 was assigned to Asian Americans in the
on the math exam, F(1, 104) ⫽ 5.94, p ⫽ .02. However, gender did ethnicity– ethnic stereotype cue condition and a weight of ⫺1 was
not interact with any of the other variables in the study. Thus, assigned to Asian Americans in the control condition and to
gender was retained in all models as a control variable. For Latinos in both conditions. I tested a second regression analysis in
descriptive purposes, the means and standard deviations for math which the interaction term for the Ethnicity ⫻ Ethnicity–Ethnic
score by ethnicity, ethnic identity cue condition, and gender are Stereotype Cue effect was replaced with the contrast code. The
reported in Table 2.1 contrast code was significant (␤ ⫽ .51, p ⫽ .005) and indicated
To test the hypotheses, I conducted a hierarchical regression that Asian Americans in the ethnicity– ethnic stereotype cue con-
analysis (Aiken & West, 1991). For this analysis, ethnic identifi- dition performed significantly better than their counterparts, thus
cation and math identification were centered at their means; gen-
der, ethnicity, and ethnicity– ethnic stereotype cue conditions were
contrast coded (gender: men ⫽ ⫺1, women ⫽ 1; ethnicity: Asian 1
There was an average of about 13 participants per cell for this test;
American ⫽ ⫺1, Latino ⫽ 1; ethnicity– ethnic stereotype cue: no thus, it is possible that no gender effects emerged because of insufficient
cue ⫽ ⫺1, cue ⫽ 1); and interaction terms were created by power. I conducted a bootstrap analysis (Efron & Tibshirani, 1993) to
multiplying together the appropriate variables. partially address this issue. Math score was regressed on ethnicity, gender,
The first analysis tested the effects of ethnicity, ethnic identifi- ethnic identity cue conditions, and the two- and three-way interactions
cation, and ethnicity– ethnic stereotype cue on math performance, between these variables using 1,000 bootstrap samples. Consistent with the
analysis of variance, the bootstrap analysis did not show any significant
controlling for gender and math identification, using a three-step
gender effects (Gender ⫻ Condition: ␤ ⫽ .08, 95% confidence interval
hierarchical regression model. In the first step, math performance [CI] [⫺.34, .55]; Gender ⫻ Ethnicity: ␤ ⫽ .01, 95% CI [⫺.45, .45];
was regressed on gender, math identification, ethnicity, ethnic Gender ⫻ Condition ⫻ Ethnicity: ␤ ⫽ .16, 95% CI [⫺.23, .62]). A more
identification, and ethnicity– ethnic stereotype cue (main effects focused test of the interaction between gender and ethnic identity cue
model). The two-way interactions between ethnicity, ethnic iden- conditions for each ethnic group also revealed no significant gender inter-
tification, and ethnicity– ethnic stereotype cue were entered on the actions.
 BRIEF REPORTS 97

Table 2 threat (Steele & Aronson, 1995) effects among Asian Americans
Hierarchical Regression Analyses Predicting Math Performance and Latinos, respectively. Consistent with previous research, re-
sults showed that Asian Americans performed better and Latinos
Math performance performed worse on a math exam when a cue implicating their
Model b SE ␤ p
ethnicity and associated ethnic group stereotypes was present, thus
replicating the stereotype boost and stereotype threat effects, re-
Model 1 (main effects) spectively. More important, as predicted, ethnic identification
Gender ⫺.43 .21 ⫺.19 .05 moderated the stereotype boost and stereotype threat effects. Spe-
Math identification .51 .23 .21 .03
cifically, high ethnically identified Asian Americans performed
Ethnicity ⫺.19 .23 ⫺.08 .40
Ethnic identification ⫺.07 .29 ⫺.02 .92 better and high ethnically identified Latinos performed worse
Ethnicity–ethnic stereotype cue .17 .21 .08 .43 when an ethnicity– ethnic stereotype cue was present. The math
Model 2 (stereotype boost–threat model) performance of Asian Americans and Latinos who did not strongly
Ethnicity ⫻ Ethnic Identification .06 .30 .02 .85 identify with their respective ethnic groups was not significantly
Ethnicity ⫻ Ethnicity–Ethnic
affected by the presence of an ethnicity– ethnic stereotype cue.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Stereotype Cue ⫺.63 .22 ⫺.28 .01

This document is copyrighted by the American Psychological Association or one of its allied publishers.

Ethnic Identification ⫻ Ethnicity/Ethnic The results are consistent with self-categorization theory
Stereotype Cue .20 .33 .07 .54 (Turner et al., 1987), which posits that group identification in-
Model 3 (moderation model) creases the likelihood that a group identity will become personally
Three-way Interaction ⫺.82 .34 ⫺.27 .02
salient and cognitively activated in the presence of situational cues
Note. Gender was coded ⫺1 (men) and 1 (women). Ethnicity was coded that implicate that identity (Spears et al., 1999). There are, how-
⫺1 (Asian American) and 1 (Latino). Ethnicity– ethnic stereotype cue was ever, other potential explanations for these results. Specifically, it
coded ⫺1 (control condition) and 1 (ethnicity– ethnic stereotype cue con- is possible that the experimental manipulation primed high and
dition). low ethnically identified individuals but that only highly identified
individuals were affected by it. For example, it is possible that
highly ethnically identified Asian Americans experienced greater
replicating the stereotype boost effect (e.g., Shih et al., 1999). To
pride and confidence in the experimental condition because of
test the expected stereotype threat effect, I assigned a weight of 3
their personal attachment to the group (E. Smith, 1993). In addi-
to Latinos in the ethnicity– ethnic stereotype cue condition and a
tion, it is possible that Latinos in the experimental condition were
weight of ⫺1 to Latinos in the control condition and to Asian
more motivated to disconfirm the negative group stereotype to
Americans in both conditions. A final regression analysis showed
maintain a positive social identity (Tajfel & Turner, 1979).
that this contrast code was significant (␤ ⫽ ⫺.45, p ⫽ .005),
Clearly, additional research is necessary to understand exactly why
indicating that Latinos in the ethnicity– ethnic stereotype cue con-
group identification increases susceptibility to the stereotype boost
dition performed significantly worse than their counterparts, thus
and stereotype threat effects.
replicating the stereotype threat effect (e.g., Steele & Aronson,
1995).
Conclusion
Step 3: Moderation Model This study showed support for the central hypothesis that group
identification increases susceptibility to the stereotype boost and
The inclusion of the three-way interaction resulted in a signif-
stereotype threat effects among Asian Americans and Latinos,
icant increase in explained variance (⌬R2 ⫽ .05), ⌬F(1, 96) ⫽
respectively. This provides initial evidence that group identifica-
5.61, p ⫽ .02. To examine the hypothesis that ethnic identification
tion moderates the stereotype boost effect and provides converging
would moderate the stereotype boost and stereotype threat effects,
evidence for Schmader’s (2002) finding that gender identification
I conducted tests of the simple slopes of ethnicity– ethnic stereo-
moderates the stereotype threat effect among women, suggesting
type cue at 1 standard deviation above and below the mean of
that group identification has similar effects across different stereo-
ethnic identification for each ethnic group (Aiken & West, 1991).
typed groups, regardless of the valance of the group stereotype
This analysis showed that for Asian Americans, the simple effect
(i.e., increases susceptibility to the stereotype boost and stereotype
of ethnicity– ethnic stereotype cue on math performance was sig-
threat effects). These findings are consistent with the theoretical
nificant for those who were high in ethnic identification (␤ ⫽ .57,
contentions of self-categorization theory and highlight an impor-
p ⫽ .01) but not for those who were low in ethnic identification
tant similarity between stereotype boost and stereotype threat,
(␤ ⫽ .07, p ⫽ .67). For Latinos, the simple effect of ethnicity–
which have been discussed as somewhat different phenomena
ethnic stereotype cue on math performance was significant for
(Wheeler & Petty, 2001). In addition, alternative explanations for
those who were high in ethnic identification (␤ ⫽ ⫺.45, p ⫽ .02)
the findings, which will need to be tested, can be derived from
but not for those who were low in ethnic identification (␤ ⫽ .65,
social identity theory (Tajfel & Turner, 1979). This suggests that
p ⫽ .63). Thus, the prediction that ethnic identification would
increase vulnerability to the stereotype boost and stereotype threat
effects was supported.2 2
I conducted a second hierarchical regression analysis to examine the
effects of ethnicity, ethnicity– ethnic stereotype cue, and ethnic identifica-
Discussion tion on number of items attempted, controlling for math identification and
gender. None of the models in this analysis were significant, suggesting
The goal of this study was to examine ethnic identification as a that the observed effects on overall math performance were not the result
moderator of the stereotype boost (Shih et al., 1999) and stereotype of differences in number of items attempted.
 98 BRIEF REPORTS

Figure 1. Moderation of stereotype boost (Asian Americans) and stereotype threat (Latinos) effects by ethnic
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

identification.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

self-categorization and social identity theories (Tajfel & Turner, Identity salience and shifts in quantitative performance. Psychological
1979; Turner et al., 1987) can provide useful frameworks for Science, 10, 80 – 83.
further examinations of the stereotype boost and stereotype threat Smith, E. (1993). Social identity and social emotions: Toward new con-
effects. ceptualizations of prejudice. In D. Mackie & D. Hamilton (Eds.), Affect,
cognition and stereotyping: Interactive processes in group perception
(pp. 297–315). San Diego, CA: Academic Press.
References Smith, J. L., & White, P. H. (2001). Development of the domain identifi-
Aiken, L. S., & West, S. G. (1991). Multiple regression: Testing and cation measure: A tool for investigating stereotype threat effects. Edu-
interpreting interactions. Thousand Oaks, CA: Sage. cational and Psychological Measurement, 61, 1040 –1057.
Armenta, B. E., & Hunt, J. S. (2009). Responding to societal devaluation: Spears, R., Doosje, B., & Ellemers, N. (1999). Commitment and the
Effects of perceived personal and group discrimination on the group context of social perception. In N. Ellemers, R. Spears, & B. Doosje
identification and self-esteem of Latinos/as. Group Processes & Inter- (Eds.), Social identity: Context, commitment, content (pp. 59 – 83). Ox-
group Relations, 12, 23–39. doi: 10.1177/ 1368430208098775 ford, England: Blackwell.
Armenta, B. E., Knight, G. P., Carlo, G., & Jacobson, R. (2008). Associ- Spencer, S. J., Steele, C. M., & Quinn, D. M. (1999). Stereotype threat and
ation between ethnic group attachment and prosocial tendencies: Me- women’s math performance. Journal of Experimental Social Psychol-
diation by cultural values. Manuscript submitted for publication. ogy, 35, 4 –28.
Dijksterhuis, A., Bargh, J. A., & Zanna, M. P. (2001). The perception- Steele, C. M. (1997). A threat in the air: How stereotypes shape intellectual
behavior expressway: Automatic effects of social perception on social identity and performance. American Psychologist, 52, 613– 629.
behavior. In Advances in experimental social psychology (Vol. 33, pp. Steele, C. M., & Aronson, J. (1995). Stereotype threat and the intellectual
1– 40). San Diego, CA: Academic Press. test performance of African Americans. Journal of Personality and
Efron, B., & Tibshirani, R. (1993). An introduction to the bootstrap. New Social Psychology, 69, 797– 811.
York: Chapman & Hall/CRC. Steele, C. M., Spencer, S. J., & Aronson, J. (2002). Contending with group
Hunt, J. S., & Espinoza, R. K. E. (2007). Prejudice against Hispanics: image: The psychology of stereotype and social identity threat. In M. P.
Analysis and development of an assessment tool. Manuscript submitted
Zanna (Ed.), Advances in experimental social psychology (pp. 379 –
for publication.
440). San Diego, CA: Academic Press.
Niemann, Y. F., Jennings, L., Rozelle, R. M., & Baxter, J. C. (1994). Use
Tajfel, H., & Turner, J. C. (1979). An integrative theory of intergroup
of free responses and cluster analysis to determine stereotypes of eight
conflict. In W. G. Austin & S. Worchel (Eds.), The social psychology of
groups. Personality and Social Psychology Bulletin, 20, 379 –390.
intergroup relations (pp. 33– 47). Monterey, CA: Brooks/Cole.
Phinney, J. S. (1992). The multigroup ethnic identity measure: A new scale
Thoman, D. B., White, P. H., Yamawaki, N., & Koishi, H. (2008).
for use with diverse groups. Journal of Adolescent Research, 7, 156 –
176. Variations of gender-math stereotype content affect women’s vulnera-
Romero, A. J., & Roberts, R. E. (2003). The impact of multiple dimensions bility to stereotype threat. Sex Roles, 58, 702–712.
of ethnic identity on discrimination and adolescents’ self-esteem. Jour- Turner, J. C., Hogg, M. A., Oakes, P. J., Reicher, S. D., & Wetherell, M. S.
nal of Applied Social Psychology, 33, 2288 –2305. (1987). Rediscovering the social group: A self-categorization theory.
Schmader, T. (2002). Gender identification moderates stereotype threat Oxford, England: Blackwell.
effects on women’s math performance. Journal of Experimental Social Umaña-Taylor, A. J., & Updegraff, K. A. (2007). Latino adolescents’
Psychology, 38, 194 –201. mental health: Exploring the interrelations among discrimination, ethnic
Shih, M., Ambady, N., Richeson, J. A., Fujita, K., & Gray, H. M. (2002). identity, cultural orientation, self-esteem, and depressive symptoms.
Stereotype performance boosts: The impact of self-relevance and the Journal of Adolescence, 30, 549 –567.
manner of stereotype activation. Journal of Personality and Social Wheeler, S. C., & Petty, R. E. (2001). The effects of stereotype activation
Psychology, 83, 638 – 647. on behavior: A review of possible mechanisms. Psychological Bulletin,
Shih, M., Pittinsky, T. L., & Ambady, N. (1999). Stereotype susceptibility: 127, 797– 826.