becoming american: stereotype threat effects in afro ... · afro-caribbean would include people...

Black immigration to the United Stateshas increased markedly since the post-1965 changes in immigration policy. As

of the 2000 census, foreign-born blacks con-stituted 12% of all first-generation immi-grants in the United States and accounted forapproximately 6% of the total U.S. black pop-ulation (Logan and Deane 2003). Althoughthere is some immigration from Africa, theprimary source of black immigration to theUnited States is from the Caribbean nations.Among Afro-Caribbean immigrants, WestIndians (defined as those countries originallycolonized by the British1) have been of partic-

Social Psychology Quarterly2007, Vol. 70, No. 4, 384–404

Becoming American: Stereotype Threat Effects in Afro-Caribbean Immigrant Groups*

KAY DEAUXGraduate Center, City University of New York

NIDA BIKMENDenison University

ALWYN GILKESMedgar Evers College, City University of New York

ANA VENTUNEAC

YVANNE JOSEPHGraduate Center, City University of New York

YASSER PAYNEUniversity of Delaware

CLAUDE STEELECenter for Advanced Study in the Behavioral Sciences

Educational and occupational data suggest that second-generation West Indian immigrantshave less favorable outcomes than their first-generation counterparts, who are typicallyshown to outperform comparison groups of African Americans. In two studies, we explore thesocial psychological process of stereotype threat as it differentially affects the performanceof first- and second-generation West Indian students. An initial questionnaire study of 270West Indian students provided data on perceived favorability of African American and WestIndian stereotypes, ethnic identification, and perceptions of discrimination. An experimentalstudy of stereotype threat showed a significant interaction between generation and stereotypethreat condition: first- and second- generation West Indian students performed equally inneutral conditions, but differed significantly when stereotype threat was present. While first-generation students increased their performance in the threat condition, second-generationstudents showed the performance decrements characteristic of African American students.Effects due to the race of the experimenter were also found, suggesting the importance ofcontext in testing situations. Overall, the findings argue for the relevance of psychologicalprocesses in understanding broader demographic patterns of immigration and change.

384

* This research was funded by the Russell SageFoundation. We are also grateful to Jennifer Eberhardt,Ewart Thomas, and Mary Waters for their early discus-sions that contributed to this research. We thank DesireeCassar for her assistance in data collection; we alsoextend thanks to Taryn Tang, Teceta Thomas Tormala,Shaun Wiley, members of the Identity Research Groupand the Immigration Research Group at the CUNYGraduate Center, and Joshua Aronson and his lab group atNew York University for their helpful comments on earli-er versions of this manuscript.

Correspondence concerning this article should beaddressed to Kay Deaux, CUNY Graduate Center, 365Fifth Avenue, New York, NY 10016; [email protected]

#3247—Social Psychology Quarterly—VOL. 70 NO. 4—70408-Deaux

1 We use the term West Indian, consistent with theusage adopted by Foner (2001), to refer to people from

BECOMING AMERICAN 385

ular interest to social scientists because of twocompeting factors. On the one hand, they arean immigrant group whose first language isEnglish, thus giving them some advantageover other immigrants who must learn a newlanguage before having ready access to educa-tional and occupational opportunities in theUnited States. On the other hand, because themajority of these immigrants are black, theyenter a country in which their skin colorbecomes the basis for discriminatory treat-ment.

To the extent that the first of these fac-tors—facility with English—is dominant, onewould predict that West Indian immigrantswould do well in educational and occupation-al domains, making progress more rapidlythan immigrants from non-Anglophone coun-tries. On the other hand, to the extent that WestIndian immigrants are subject to color-baseddiscrimination, they should be impeded rela-tive to white immigrants. Indeed, one mightexpect that being subjected to some of thesame discriminatory conditions that confrontnative-born African Americans, West Indianswould become similar to these groups in avariety of psychological ways. Of particularinterest here is the degree to which WestIndian immigrants are susceptible to theeffects of stereotype threat, in which an aware-ness of negative group stereotypes about thecapability of African Americans results in per-formance decrements for members of thisgroup (Steele 1997; Steele and Aronson 1995;Steele, Spencer, and Aronson 2002).

BACKGROUND AND THEORY

Occupational and Educational Outcomes ofWest Indian Immigrants

Despite the potential for negative treat-ment, West Indian immigrants have often

fared quite well by traditional educational andoccupational markers. Typically these assess-ments are made in comparison to native-bornAfrican Americans. Model (1991, 1995),using census data from 1970, 1980, and 1990,found that West Indian men and women havehigher labor force participation rates thannative-born African Americans. She alsoreported that first- generation West Indianimmigrants show higher educational aspira-tions and greater educational attainment thannative-born African Americans. Similarly,Kalmijn (1996) found that black immigrantswere more educated, had higher prestige occu-pations, and earned more than blackAmericans. Although there continues to besome debate as to whether West Indians havean earnings advantage over native-bornAfrican Americans, the employment andoccupational advantage of first-generationWest Indians appears to be reliable (Waters1999b).

Whereas first-generation West Indianimmigrants systematically outperform native-born African Americans, their advantage isnot as clear in the second generation. Waters etal. (1998) found that second-generation WestIndian respondents typically report manyspells of short employment. Additionally, theeducational achievement of second-generationWest Indians often declines from elementaryschool to high school. .It should be noted thatthis decline from first to second generation isin contrast to the linear process of assimilationthat characterized earlier generations of whiteimmigrant groups (Portes 1995). Indeed, con-temporary analyses of immigrant pathwaysfrom first to second generation (often referredto as models of segmented assimilation) pointto downward assimilation as one possible out-come, most likely to be experienced by groupswho are most subject to societal discrimina-tion (Portes 1995; Portes and Rumbaut 2001).

Sociological theories that attempt toexplain West Indian success focus on factorssuch as: the selectivity effects of immigration(Portes and Rumbaut 1992), the developmentof social and community networks (Portes1995; Tilly 1990), a preference by whites forforeign-born blacks (Kasinitz and Rosenberg1996; Waters 1999b), and cultural differences


English-speaking islands of the Caribbean, including themainland nations of Belize and Guyana. For our purpos-es, it was important to restrict participation to those forwhom English was their first language so that first- andsecond-generation participants could be assumed to beequivalent in English proficiency. The more general termAfro-Caribbean would include people from countries inwhich another language, such as French or Spanish,would typically be the first language.

386 SOCIAL PSYCHOLOGY QUARTERLY

between West Indians and black Americans(Reid 1939). Although supportive data areavailable for some of these explanations, moreoften these factors have been suggested aspossible causes but not tested empirically.More importantly for the present purposes,although some of these theories work reason-ably well in explaining first-generation pat-terns, they typically do not speak to possibledifferences between the generations. Wouldwe expect, for example, that immigrants whoarrive as early teenagers with their parentswould be any more or less highly selected thanchildren born in the United States to parentswho also chose to come to the United States?Or would one predict that first- and second-generation black immigrants would have dif-ferent experiences with race discrimination inthe United States? What is needed, we argue,is the introduction of a psychological perspec-tive that can more closely look at the individ-ual-level processes that may be critical to theimmigrant experience.

The Potential Role of Stereotype Threat

Work on stereotype threat by Steele andhis colleagues (Aronson et al. 1999; Steele1997; Steele, Spencer, and Aronson 2002)compellingly shows that when negativestereotypes about a group’s abilities andpotential are “in the air,” they can underminethe performance of members of that group.When a stereotype is believed to be relevant toa domain of performance, it poses the threatthat the person will be judged or treated interms of the stereotype. The impact of thatthreat is reduced performance on domain-rel-evant tasks, an effect that has been consistent-ly demonstrated in scores of studies acrossgroups varying in gender, ethnicity, and socialclass.

Particularly relevant to the present workare studies that consider the performance ofAfrican American students (e.g., Steele andAronson 1995). Here the data show that whenthe negative stereotype of black intellectualability is made salient (for example, by char-acterizing the test as diagnostic vs. non-diag-nostic), African American students performmore poorly on achievement tasks than do

whites. In contrast, when ethnicity is notsalient, the two groups perform at equivalentlevels (after adjusting for SAT scores). Theseexperimental data are important in arguingagainst explanations that rely primarily onassumed differences in cultural dispositions.African Americans and whites can differ intheir academic performance, but whether theydo or do not in this particular task settingdepends on characteristics of the situation towhich they were randomly assigned, not toinherent differences in their capabilities or tothe cultural context and social structure thatframe their lives.

In addition to the experimental demon-strations of stereotype threat provided bySteele and his colleagues, Massey and his col-leagues (Massey et al. 2003; Massey andFischer 2005) have shown evidence of thephenomenon in a large-scale survey study ofseveral thousand first-year college students at28 selective academic institutions in theUnited States. They found that the psycholog-ical state of stereotype threat, conceptualizedin their terms as the combined effects ofdisidentification and performance burden, ledto lower grade point averages for both blackand Latino students. In their calculations, thedifference between maximum and minimumlevels of the stereotype threat factors that theyassessed meant a difference of approximatelyone-third of a point in grade point average ofblack students in their first year of college(Massey and Fischer 2005).

In the present study, we draw from thesedata on the impact of stereotype threat onAfrican Americans to explore the implicationsfor first- and second-generation black immi-grants from English-speaking countries of theCaribbean. Specifically, we ask whether thedifferences between first- and second-genera-tion West Indians suggested by some of theeducational and occupational outcome datamight be in part explained by a greater sus-ceptibility to stereotype threat on the part ofthe second generation. With more time in theUnited States, second- generation black immi-grants would have had more exposure to thenegative stereotypes about black intellectualcapability and those stereotypes might be



more likely to be salient when they themselveswere engaged in intellectual tasks.

This prediction assumes, of course, thatthere are distinctive stereotypes about the twogroups and that evaluations of West Indiansare more favorable than those of AfricanAmericans. There is some support for thisassumption. In interviews with employers,Waters (1999a) asked about perceived differ-ences between foreign-born and American-born blacks. White employers frequentlyreported differences, most typically seeingWest Indians as more ambitious and hardworking and African Americans as more trou-blesome. In a study with diverse college stu-dents, Tormala (2005) asked participants tolist positive and negative stereotypes about“Blacks in the United States” and “Blackimmigrants to the United States,” as theythought these would be endorsed by “mostpeople in the United States.” Her respondentsgenerated significantly more negative thanpositive stereotypes about black Americans; incontrast, there was no difference in the fre-quency of positive and negative stereotypesassigned to black immigrants. Qualitatively,her analysis showed that stereotypes of blackimmigrants are far less consistent than stereo-types of native-born African Americans.(However, the category itself is probably morediverse in this case, potentially includingimmigrants from Africa as well as theCaribbean.)

The existence of differential stereotypesin the culture at large does not, of course,establish that West Indians themselves areaware of these stereotypes. However, data bothfrom the West Indian workers that Waters(1999a) interviewed and from the black immi-grant participants in the Tormala (2005) studysuggest that stereotypes about blackAmericans are both known and endorsed byWest Indians themselves. Although these stud-ies give us some basis for our assumption thatthere is an awareness of negative stereotypesabout black Americans, it will nonetheless beimportant that we establish the existence ofthose perceptions in the particular populationthat we are tapping. Further, it is important toknow not only what our participants believe tobe the stereotype about African Americans,

but also what they believe is the prevalentstereotype about their own group of WestIndian origin. As shown by recent sociologicalwork (Sigelman and Tuch 1997; Torres andCharles 2004), these metastereotypes—beliefsabout the stereotypes of others toward one’sown group—can influence behavior andindeed are a basis for the operation of stereo-type threat.

To the extent that West Indian immigrantsbelieve that there is a meaningful distinctionbetween their group and the more generalblack American group, and in particular a dif-ference that would favor their ingroup over theoutgroup, we might also anticipate that theycould show enhanced performance under con-ditions of stereotype threat. This stereotype lifteffect, as documented by Walton and Cohen(2003), refers to a boost in performanceshown by members of groups (e.g., whites,males) who are not negatively stereotypedthemselves but are aware of the negativestereotypes associated with comparisongroups. Although stereotype lift effects aretypically not as pronounced as stereotypethreat effects (Walton and Cohen 2003), theyare more apt to emerge when people believe inthe validity of the negative stereotypes. In thepresent case, a stereotype lift effect would bemore likely if West Indians assumed that theirethnic group was more favorably regardedthan African Americans and wanted to distin-guish themselves from that group.

Ethnic Identification in West Indians

Earlier investigators typically assumedthat one’s ethnic identification was cotermi-nous with the categorical definition, that is, ifyou were born of Italian parents, then yourethnic identification was as an Italian. Morerecently, however, most social psychologistshave recognized that ethnic identification is asubjective state as well as, and perhaps moreimportantly than, an objective characterization(Ashmore, Deaux, and McLaughlin-Volpe2004; Deaux 1996, 2006; Waters 1990). Likeother forms of social identification, it is essen-tial to consider what people call themselves, alabeling process that can vary both in the cat-egory that is claimed and the importance of



that category (Deaux 1996, 2006). In the caseof ethnic identification specifically, Waters(1990) found, through an analysis of censusdata as well as interviews with ethnic whitesin the United States, that ethnic identificationis a personal choice, both in terms of whichethnic identity to claim and whether to self-describe in terms of ethnicity at all. Further,Waters discussed the ways in which ethnicidentity can change over the life course, bothin its importance and the meanings it holds.

In the case of West Indian immigrants,Waters (1994, 1999a, 1999b) has also report-ed on the variability in ethnic labeling. Ininterviews with adolescent West Indian immi-grants in New York City, Waters found threedifferent patterns of identification. Of hersample, 31% identified as West Indian (eithergenerally, or with specific reference to theircountry of origin). Somewhat more of theseadolescents (41%) identified primarily asAfrican American, having shifted from thecountry of origin to the immediate context asa source of identification. A third group, rep-resenting 27% of her sample and consistingprimarily of people who had more recentlyimmigrated to the United States, thought ofthemselves in terms of a more generic immi-grant category, rather than a nationality-linkedidentity. Thus, although the demographicbackgrounds of these youth were quite similar,their subjective definitions of ethnicityshowed considerable variation. It is importantto note, however, that the current conditions ofthese three groups of immigrant youth werenot always equivalent. Residential neighbor-hood and quality of school both variedbygroup, with those who considered themselvesAfrican American more likely to be in poorerneighborhoods and lower-quality schools.

The potential for variations in ethnic iden-tification between first- and second-genera-tion immigrants seemed to us considerable,even with age held constant, and presentedimportant implications for the predicted gen-erational effects in performance under stereo-type threat. If a West Indian immigrant comesto identify primarily as African American, asdid 41% of the adolescents in Waters’ (1994)sample, then would they be more susceptibleto stereotype threat than would those immi-

grants who continue to identify with theircountry of origin, where negative stereotypesabout Blacks are not part of the cultural repre-sentations? Within the stereotype threat litera-ture, the role of group identification is nottotally clear. Schmader (2002), focusing onwomen’s math performance, found thatwomen with stronger gender identificationshowed greater decrements in performanceunder stereotype threat conditions than didwomen for whom gender identification wasnot as strong. With regard to ethnic identifica-tion among African Americans, however, thedata are less consistent (Steele et al. 2002).2

Nonetheless, it seems reasonable to hypothe-size that immigrants who identify as AfricanAmericans would be more likely to see stereo-types about African Americans as being rele-vant to themselves than would immigrantswho continue to identify with their country oforigin. However, in line with the Steele andAronson (1995) findings, we would expect tofind differences in performance only when theconditions made these stereotypes salient.

Another potential moderator of stereotypethreat effects in West Indian immigrants isexperience with race-based discrimination.Would more experience with discriminationlead, not only to stronger expectations of andanxiety about being discriminated against, butalso greater susceptibility to threats related tostereotypes of African Americans? Or wouldthe prevalence of race as a master status in theUnited States override any distinctionsbetween generational cohorts?

To summarize, numerous sociological anddemographic studies of West Indian immigra-tion have concluded that first-generation WestIndians are likely to succeed in educationaland occupational domains at rates greater thantheir native-born African American cohorts.


2 One explanation for some of the inconsistent datawith regard to ethnic identification among AfricanAmericans, as discussed by Steele et al. (2002), mayinvolve differences between simple scalar measures of theimportance of an identity on the one hand, and more qual-itative meaning systems associated with an identity (thelatter of which could include both domain relevance andcoping strategies). See Ashmore, Deaux, andMcLaughlin-Volpe for a more extensive discussion of thevarious dimensions of collective identification.


Many of the proffered explanations for theseobservations come up short, however, whenfaced with evidence of second-generationdecline. Our goal in this research is to explorethe viability of social psychological explana-tions for the generational shift that has beenobserved, and in so doing to highlight theways in which context and psychologicalprocess interact.

Hypotheses

Reasoning from the available evidenceand theory, we hypothesized that the differ-ences between first- and second-generationWest Indian immigrants are associated, at leastin part, with differential susceptibility tostereotype threat effects. Specifically, we pre-dicted that first-generation West Indian immi-grants are protected from or insensitive tostereotype threat, and thus will not show a per-formance decrement when stereotypes aresalient, that is, when a test is described asdiagnostic of ability.

In contrast, we predicted that second-gen-eration West Indian immigrants will show thesame pattern of stereotype threat effects typi-cally evidenced in African American respon-dents, that is, decrements in performancewhen a test is labeled diagnostic as opposed tonon-diagnostic. Thus, we predicted a signifi-cant interaction between stereotype threatcondition and generation. Potential modera-tors of this predicted effect, which might beexpected to vary between first- and second-generation immigrants include: (a) ethnicidentification, (b) metastereotypes of AfricanAmericans and West Indians, and (c) expecta-tions of race-based discrimination.

The study was conducted in two parts.First, a questionnaire was administered to alarge sample of West Indian immigrants, bothfirst and second generation, all of whom werestudents within a large public university sys-tem in New York City. Second, a subsample ofthe initial group was contacted to participatein an experimental study in which the condi-tions of stereotype threat were manipulatedand performance on an achievement test wasassessed.

Although a primary purpose of the ques-tionnaire study was to identify students to par-ticipate in the subsequent experimental ses-sion, data collected in the questionnaire wereimportant in several ways. First, and perhapsmost important, questions about perceivedstereotypes of both African Americans andWest Indians could be evaluated to determinewhether first- and second- generation WestIndians differ in their perceptions of the posi-tivity or negativity of these stereotypes. If sub-sequent experimental results show a resistanceto stereotype threat effects, it will be impor-tant to know if such resistance is based on alack of awareness of the stereotypes, or if itreflects knowledge of the stereotypes but abelief that those stereotypes are irrelevant tothe self. Second, comparisons of first- andsecond-generation students in terms of ethnicidentification and sensitivity to discriminationallow us to further characterize the psycholog-ical similarities or differences between the twogenerations.

QUESTIONNAIRE STUDY

Method

Participants. New York City has been a majordestination for immigrants from the Caribbean,particularly since the 1965 changes in U.S.immigration policy. They are now the largestimmigrant group in the city, with approxi-mately half a million first-generation WestIndian immigrants living in the city by the late1990s (Foner 2005). At the same time, unlikesome other immigrant destinations such asLondon, Black immigrants to New York entera community that has a large native Blackpopulation. In 2000, the black population ofNew York City was approximately 25% of thecity’s total population of 8 million; 28% of thisproportion was foreign-born (Foner 2005).Thus questions of race and ethnicity are exam-ined and defined within a multicultural con-text.

In this study, a total of 270 West Indianstudents who were currently enrolled at one ofthe 4-year undergraduate colleges within theCity University of New York system complet-ed the initial questionnaire. Students identi-fied through university records as being of



West Indian heritage were contacted by letters;additionally, the researchers made direct con-tact with students at three selected collegeswithin the system that have a high percentageof enrolled West Indian students. All studentswho agreed to complete the questionnairewere paid $10 for their participation.

Questionnaire. The questionnaire wasdesigned to assess a number of concepts perti-nent to the hypotheses of the study, as well asrelevant background material on the partici-pants. Key concepts in terms of the theoreticalframework that we are using were stereotypeknowledge and ethnic identification.Additional measures collected to explore pos-sible differences between groups includedsensitivity to race-based rejection and demo-graphic material, including immigration histo-ry.

Favorability of African American and WestIndian Stereotypes. Perceived favorability ofcultural stereotypes was assessed with twoscales, one with African Americans as the tar-get group and the other with West Indians asthe target group (with the order of presenta-tion of these two scales counterbalanced). Ineach case, respondents were asked to rate (ona 6-point scale) the degree to which each of 12adjectives (6 positive and 6 negative) werebelieved by people in general to be character-istic of the target group. Positive adjectivesincluded traits such as hard-working, smart,and friendly; negative adjectives includedpoor, lazy, and criminal. Ratings on the nega-tive items were reverse coded, and the 12items were summed for an overall positivityscore.

Ethnic identification. Ethnic identificationwas assessed in two ways. First, we usedPhinney’s Multi-Ethnic Identity measure(1998), which includes an initial open-endedquestion of ethnic identification followed by a12-item scale. A second measure of ethnicidentification was developed to more directlyassess relative identification with West Indianversus African American ethnicity. This mea-sure consisted of a single 5-point scale that

asked respondents to indicate whether theyconsidered themselves “definitely WestIndian,” “more West Indian than AfricanAmerican,” “equally West Indian and AfricanAmerican,” “more African American thanWest Indian,” and “definitely AfricanAmerican,” where 1 indicated the strongestWest Indian identification and 5 the strongestAfrican American identification.

Luhtanen and Crocker’s (1992) measureof collective self-esteem was also included asan additional measure of ethnic identification.We used three of the four subscales from thismeasure: identity (the importance of the groupto one’s self concept), private regard (thefavorability that one has toward his or her owngroup), and public regard (the perceivedfavorability of others toward one’s group).Each subscale consists of 4 items, to whichparticipants respond on a 7-point scale rang-ing from strongly disagree (1) to stronglyagree (7). A group-specific version of the CSEwas used, such that each statement was madein reference to the ethnic group that therespondent had chosen in the open-endedquestion of the Multi-Ethnic Identity measure.

Expectations of discrimination. A shortenedversion of the Sensitivity to Race-BasedRejection measure (Mendoza-Denton etal.2002) was included in order to assess possi-ble differences between first- and second-gen-eration immigrants in their anxiety and expec-tation of discrimination. Participants wereasked to imagine themselves in each of six sit-uations (e.g., you are in a store and a clerkglances your way). In each case, participantsare asked to estimate the degree to which theywould expect that their race/ethnicity wouldbe an issue (e.g. “The clerk might continue tolook at me because of my race/ethnicity”) andhow anxious they would be about that situa-tion. Separate scores for expectation and anx-iety were calculated, each summed over thesix situations.

Demographic information A variety of demo-graphic questions were included to assessimmigration history of the respondent andhis/her parents, including country of birth,



date of immigration to the United States, par-ents’ occupations and education, and their per-ceived class standing. Also included in thissection of the questionnaire were inquiriesabout the degree of contact the respondent hadwith their Caribbean origin and their use ofcultural media related to those countries.

Results

Sample CharÏacteristics. Of the total 270 par-ticipants, 145 were first-generation immi-grants (born outside of the United States) and125 were second generation (born in theUnited States to parents who were born in theWest Indies). In this sample, 61% were femaleand 39 % were male. The average age of stu-dents in the sample was 22.3; first-generationstudents were slightly older on average (M =23.4, SD = 6.6) than were second-generationstudents (M = 21.0, SD = 3.2), t(266) = 3.76,p < .001.

In terms of the ethnic identification of thegroup, the sample was clearly skewed towardsidentification as West Indian. On the 5-pointscale where 1 = definitely West Indian (WI),3 = equally West Indian and African Ameri-can, and 5 = definitely African American(AA), the mean score of the sample was 2.2(SD = 1.1). In percentage terms, 63% of the

sample considered themselves definitely orprimarily West Indian; 25% rated themselvesequally West Indian and African American,and only 12% regarded themselves as defi-nitely or primarily African American.

Comparison of First- and Second-GenerationImmigrants. Means and standard deviationsfor the two groups are shown in Table 1.Differences between the two generationalgroups appeared on a number of measures, thetwo most interesting for our purposes beingmeasures of ethnic identification and per-ceived stereotypes and evaluations of WestIndians and of African Americans.

First, with regard to ethnic identification,first-generation students more strongly identi-fied as West Indian than did second-genera-tion students. This is evident in the significantdifference in mean scores on the 5-point iden-tification scale, t (260) = 6.35, p < .001, aswell as in a significant correlation betweengeneration and ethnic identification, r (262) =.37, p < .001. It is important to note, however,that these differences are relative rather thanabsolute. Mean scores on ethnic identificationare to the West Indian side of the midpoint forboth first-generation (M = 1.8, SD = 1.0) andsecond-generation (M = 2.6, SD = 1.2) stu-


Table 1. Comparison of First and Second Generation Afro-Caribbean Immigrants

Generation

First (n = 145) Second (n = 125)M(SD) M(SD) t-value

Ethnic identification 01.80 (1.0) 02.60 (1.2) 6.35**Collective self-esteem:—Identity importance 19.00 (5.1) 19.00 (5.5) 0.06—Private regard 25.70 (3.7) 25.00 (3.9) 1.66—Public regard 19.90 (4.7) 17.10 (5.2) 4.52**Multi-Ethnic Identity Measure:—Search 25.60 (5.6) 24.80 (5.3) 1.15—Belonging 43.80 (6.0) 41.50 (5.9) 3.08**Sensitivity to race-based rejection: 13.40 (7.3) 13.40 (7.1) 0.05—Race-based anxiety 21.70 (7.0) 22.40 (7.2) 0.75—Race-based expectations 20.70 (6.7) 20.70 (6.4) 0.04Probability of returning to live in the 03.40 (1.4) 02.30 (1.4) 6.65**—West Indies at some future dateClass-relevant data:—Mother’s educationa 03.15 (1.41) 02.50 (1.20) 3.86**—Father’s educationa 03.09 (1.45) 02.99 (1.31) 0.527—Perceived class standing 02.32 (1.07) 02.89 (0.97) –4.47**

Notes: * p < .05; ** p < .01a scale from 1 (some graduate work) to 5 (did not complete high school).


dents, indicating that although these second-generation students are shifting toward anincorporation of African American into theiridentity definition, they continue to maintain astrong sense of West Indian identification aswell.

Although first- and second-generationstudents differed in their endorsement of spe-cific ethnic labels, the strength of their ethnicidentification was for the most part equiva-lent. On both the Identity and Private Regardsubscales of the collective self-esteem scale,in which the referent was whatever group labelthey chose as self-identifying, first- and sec-ond-generation students did not differ. That is,they accorded their ethnic identificationequivalent importance and equally regardedthat identity as positive. The groups alsoscored similarly on the Search subscale of thePhinney measure. On the belonging subscaleof the Phinney measure there was a significantdifference between groups, with first genera-tion students showing stronger endorsementthan second-generation students, t (265) =3.27, p < .001. In addition, the groups differedsignificantly in their scores on the PublicRegard subscale of the collective self-esteemscale, a scale that assesses how one thinksother people regard their group. In this case,first-generation students believed people weremore favorable to their group (M = 19.9, SD =4.7) than did second-generation students (M =17.1, SD = 5.2), t (262) = 4.52, p < .001.

This difference in the perceptions of oth-ers’ evaluations is also reflected in the mea-sures of stereotyping, in which we assessedthe favorability of perceived stereotypes ofboth African Americans and West Indians. Asshown in Table 2, first-and second generationstudents did not differ in their metastereotypes

of African Americans, but did differ signifi-cantly in their metastereotypes of WestIndians, with the first generation having amore favorable estimate of others’ views. Weexplored this difference further with a 2 � 2analysis of variance with generation as abetween-subjects factor and ratings of ethnicgroups a within-subjects factor. This compari-son yielded main effects both for ethnic group,F (1,247) = 254.7, p < .001, �2 = .51, and forgenerational status, F (1,247) = 10.8, p < .001,�2 = .04. All students believed that WestIndians are regarded more favorably than areAfrican Americans, and first-generation stu-dents gave more favorable ratings to bothgroups than did second-generation students.These main effects were qualified by an inter-action between generation and ethnicity of thetarget group, F(1,247) = 5.7, p < .02, �2 = .02.Among the second generation, the perceivedadvantage of West Indians is less, droppingapproximately five points. (There is no corre-sponding increase in the perceived favorabili-ty of African Americans, which is essentiallyunchanged.)

Interestingly, first- and second-generationimmigrants did not differ from each other onthe measure of Sensitivity to Race-basedRejection. On both the anxiety and expecta-tion scales of the measure, the two groupsexpressed equivalent reactions.

Comparisons of the two groups on thevarious demographic questions showed a fewscattered differences between first- and sec-ond-generation students. The mothers of first-generation students had less education, t (260)= 3.94, p < .001, and first-generation studentsperceived their parents to be lower in socioe-conomic standing, t (240) = 4.55, p < .001,both as compared to the second-generation


Table 2. Favorability of Metastereotypes

Generation

First (n = 145) Second (n = 125)M(SD) M(SD) t-value

Metastereotypes of: —African Americans 35.3 33.8 1.17

(9.8) (9.6)—West Indians 47.8 42.7 4.13**

(9.6) (10.1)

Note: * p < .05; ** p < .01


students. Firstgeneration students alsobelieved it more likely that they might returnto live in the West Indies at some future datethan did second- generation students, t (263) =6.64, p < .001.

Men and women differed only in terms ofperceptions of discrimination. Men were morelikely to expect race-based rejection thanwomen were (p < .01) and they tended to bemore anxious about such rejection (p = .07).There were no other gender differences in anyof the study variables.

Discussion

In order to hypothesize stereotype threateffects, we needed to establish that WestIndian immigrants have knowledge of theunfavorable image of African Americans andthat first- and second-generation immigrantsdo not differ in that regard. The data show thatto be the case: both first- and second-genera-tion West Indians believe that stereotypes ofAfrican Americans are less positive thanstereotypes of West Indians, and the two gen-erations do not differ in their metastereotypeof African Americans. Thus, the beliefs are “inthe air,” as Steele (1997) observed, and that airis the same for both generations of WestIndian immigrants.

What does differ between the generationsis the metastereotype of West Indians. For sec-ond-generation immigrants, the favorability ofthis image has decreased–not to the level ofthe African American stereotype, but substan-tially below the level endorsed by first-gener-ation immigrants. This more negativemetastereotype of West Indians is reflectedalso in the public regard measure of collectiveself-esteem. Here we found a significant dif-ference between first- and second-generationimmigrants, such that the latter believed thepublic had a less favorable view of their groupthan did first-generation students.

The question to be addressed is whetherthe negative beliefs about African Americansconstitute an equivalent threat for first-andsecond-generation immigrants. We arehypothesizing that they do not. First-genera-tion immigrants, although aware of the stereo-types associated with African Americans, are

hypothesized not to see those images as self-relevant to the same degree that second-gener-ation immigrants do. Because first-generationimmigrants have spent some years in aCaribbean country, where Blacks are in themajority and race discrimination is not preva-lent, we assume that they are able to maintainsome distance between society’s unfavorableviews of African Americans and their imageof themselves as a different kind of black.Second-generation immigrants, in contrast, donot have that same experience. Rather, frombirth onward, they are often treated by othersas members of the category African American,not distinguishably different from native-bornAfrican Americans. Thus, we hypothesize thatsecond-generation students will show thesame type of stereotype threat effects thathave been previously shown in AfricanAmerican populations. In contrast, we expectthat first-generation immigrants will show noperformance decrement when stereotypethreat conditions are present.

As suggested earlier, subjective ethnicidentification may also influence the degree towhich Afro-Caribbean immigrants are suscep-tible to stereotype threat effects. Comparisonsof first- and second-generation immigrants inour questionnaire data establish some impor-tant parameters. First, the data show that eth-nic identification, as assessed both by theidentity importance and the private regardsubscale of the Collective Self-Esteem scale,is equivalent for the two groups. Thus, what-ever group they are identified with is equallyimportant to their self-definition. However,the definition of that group differs significant-ly between the two groups, as the measure ofethnic identification (a dimension rangingfrom “definitely West Indian” to “definitelyAfrican American”) indicates. First-genera-tion immigrants were significantly stronger intheir endorsement of a West Indian identity. Atthe same time, it is important to note that themajority of all the participants placed them-selves toward the West Indian side of thescale, showing a more skewed distributionthan Waters (1994) observed in her highschool sample. The difference between thetwo samples could be due to differential selec-tion based on socioeconomic factors. Waters



(1994) reported that ethnic identity was asso-ciated with socioeconomic status in her sam-ple, such that the students who identified asAfrican American were more likely to be inpoorer neighborhoods with lower qualityschools. Those students, we can surmise,would be less likely to enter college thanwould students from better school systemsand neighborhoods, students who in theWaters study were more likely to hold on to aWest Indian identity. Thus, our sample of col-lege students would be biased in favor of thosemore academically prepared students who aremore identified with their West Indian her-itage.

Finally, it is worth noting that despite dif-ferences in identification and length of time inthe country, the two generations do not differin their expectations of and anxiety about pos-sible discriminatory experiences. Althoughone might assume that first-generation stu-dents had less actual experience with discrim-ination directed toward them, the approxi-mately 10 years that they had spent in theUnited States was apparently ample time forthem to expect and be anxious about the pos-sibility of being a target. These findings reaf-firm the widely-accepted idea that being blackis a “master status” within the United States(Foner 2005), a category used by others thatcommonly ignores possible distinctionsamong people who vary in ethnicity andnationality.

Our questionnaire data establish the dif-ference between first- and second-generationimmigrants in terms of their ethnic identifica-tion, showing a greater or lesser tendency,respectively, to identify as West Indian.Although this covariance is far from perfect, itdoes establish a psychological differencebetween the two demographic categories thatis consistent with our theoretical position.Equally important is the evidence that bothfirst- and second-generation immigrants havea clear understanding of the unfavorablestereotype that exists in U.S. society withregard to African Americans. Thus any evi-dence that first-generation immigrants are notsusceptible to stereotype threat effects wouldnot be based on a lack of awareness of the

stereotype, but rather some protection or dis-tancing from that stereotype.

The purpose of the experimental study isto test the hypothesis that second-generationWest Indian immigrants are more susceptibleto stereotype threat than are first-generationimmigrants. In addition, we consider theextent to which possible moderators, such asethnic identification, might influence reac-tions to stereotype threat.

An additional factor in the experimentaldesign is the race of the experimenter, a vari-able that has been shown to be influential insome previous research (e.g., Danso and Esses2001; Katz and Greenbaum 1963; Katz,Roberts, and Robinson 1965). Katz et al.(1965), for example, found that the presenceof a white experimenter caused decrements inthe performance of black students when thetask was presented as an intelligence test,while it enhanced performance when the sametask was presented as a motor coordinationtask. If we assume that the intelligence testinstructions are comparable to stereotypethreat conditions, then we might expect thatdecrements would be greater for second-gen-eration students when the experimenter waswhite as opposed to black. Support for thisprediction also comes from Massey andFischer’s (2005) study. Students whose collegeexperience included a higher than averagenumber of minority faculty did not report thekind of performance burden that was associat-ed with lower academic performance. In linewith these studies, we anticipated a possiblethree-way interaction, namely that the predict-ed two-way interaction of generation andstereotype threat condition would be morelikely to occur with white experimenters andless likely, or even absent, in the presence ofBlack experimenters.

EXPERIMENTAL STUDY

Method

To assess the effect of stereotype threat onthe performance of West Indian immigrants,we conducted an experimental study in whichfirst- and second-generation students wererandomly assigned to conditions that would orwould not activate stereotype threat. The goal



was to determine whether, within this specificsample, performance would vary as a functionof the particular experimental condition. A2 � 2 � 2 design was used, crossing genera-tion of student (first vs. second), diagnosticityof the test (diagnostic vs. non-diagnostic ofability) and race of experimenter (black vs.white).

Participants

From the sample of students who com-pleted the questionnaire, 75 students (41women and 34 men) were recruited to partici-pate in the experimental study of stereotypethreat. First- and second-generation studentswere recruited in approximately equal num-bers (N = 41 and 34, respectively). Becauseself-ascribed ethnic identification was corre-lated with generation, and because the distrib-ution of ethnic identification scores was heav-ily skewed toward identification as WestIndian, it was not possible to select partici-pants on the basis of their identification, inde-pendent of generational status. However, wedid attempt to recruit as many students as pos-sible from that group who identified them-selves as more African American or at leastequally West Indian and African American tobalance out the more numerous West Indianidentified participants.

Procedures

Participants were tested in small rooms atthe colleges, in groups ranging in size fromone to six. Typically two experimenters werepresent at a session, and we kept teams “colorconsistent”, thus allowing us to systematicallyconsider possible experimenter effects.3 Raceof experimenter team was randomly assignedacross experimental conditions. Both first-and second-generation students were mixedwithin a typical session, and the experimenters

were not aware of the generational status ofthe participants in their sessions.

Manipulation of stereotype threat. One exper-imenter read the instructions aloud to partici-pants, and participants could read along withthe same text, printed on the first page of theirbooklet. The manipulation of stereotype threatwas contained within these instructions, usinga task description that has proved to be effec-tive in establishing different levels of stereo-type threat (cf. Steele and Aronson 1995). Inthe Diagnostic (Stereotype Threat) Condition,the instructions stressed that the test was anassessment of the student’s verbal abilities andlimitations. In the Non-diagnostic condition,the test was described as an exercise in testdevelopment, evaluating the test itself ratherthan individual ability. The assumption madeby Steele and his colleagues, and supported byconsiderable research, is that this seeminglysubtle salience of diagnosticity of the test willbring the group stereotype to mind, along withthe threat of confirming that stereotype(Steele, Spencer, and Aronson 2002).Participants were then given 25 minutes toanswer as many questions as they could, on a27-item exam. Items were taken from a GREpreparation text and were selected to be rea-sonably difficult for the study sample. Theaverage performance was 9.8 out of a possiblescore of 27, thus verifying the anticipated dif-ficulty of the test.

At the end of the allotted time period, theexperimenters collected the performancematerial and participants were given a postex-perimental questionnaire to complete, assess-ing their views of the test, their performance,and the testing conditions. Participants werethen debriefed, providing them a full explana-tion of the ways in which instructions affectperformance and of the normative difficultyof the test. The participants were asked not todiscuss the findings with others at their col-lege until the end of the term.

Results

The principal test of our hypothesis isbased on a 2 � 2 � 2 analysis of variance,appropriate to the experimental design that we


3 A team of three white (all female) and three black (2male, 1 female) experimenters shared the experimenterrole. One of the black experimenters was a West Indianimmigrant himself. In order to control for the potentialinfluence of his accent, he assumed a non-speaking rolewhen paired with one of the other experimenters.


used and consistent with other studies ofstereotype threat (e.g., Steele and Aronson1995). The three dichotomous categories,experimentally manipulated or controlled,were generation (first vs. second), diagnosticcondition (threat vs. no threat), and race ofexperimenter (black or white), with perfor-mance as the dependent variable. Performanceon the test can be assessed in three differentways: in terms of number attempted, numbercorrect, and percentage correct. In the resultsdescribed below, we use percentage correct asthe primary dependent variable, consistentwith a number of previous investigators. Ingeneral, using the simple number correct pro-duces similar results, while the number ofitems attempted does not vary by experimen-tal condition.

Tests of Generational Hypotheses

Analysis of variance results are presentedin Table 3. This analysis showed no main

effects of diagnostic condition, generation, orexperimenter team on the measure of percent-age of problems solved correctly. The predict-ed two-way interaction between generationand diagnostic condition on performance wassignificant, as shown in Table 3 and in Figure1.

Tests of simple effects within each diag-nostic condition showed that the differencebetween first- and second-generation partici-pants was not significant when the test waspresumed to be non-diagnostic. This lack of adifference is theoretically important, as itcounters claims that first- and second-genera-tion immigrants possess intrinsic differencesin motivation, ability, or cultural characteris-tics that might account for differential out-comes. Instead, we find that when conditionsare psychologically neutral, there is virtuallyno difference in capability on a difficult acad-emic test.


Table 3. Analysis of Variance of Performance Data (Percentage correct)

F p �2

Generational status 0.31 0.58 0.005Stereotype threat condition 0.22 0.64 0.003Experimenter’s race 0.01 0.92 0

Generation � Condition 4.59 0.036 0.064Generation � Experimenter 4.09 0.047 0.057Condition � Experimenter 12.55 0.001 0.158Generation � Experimenter � Condition 1.21 0.275 0.018

Figure 1. Performance (Percentage Correct) of First- and Second-Generation Students in Diagnostic and Non-Diagnostic Conditions.


In contrast, the performance of first- andsecond-generation students differed signifi-cantly when the instructions stressed the diag-nosticity of the test, t (35) = 1.95, p < .05, one-tailed.4 It should be noted that changes in theperformance of both groups contribute to thissignificant effect: first-generation studentsperform better when diagnosticity is stressedthan they do in more neutral conditions, whilesecond-generation students show the antici-pated decrement in performance.

To further explore the significant interac-tion between generation and diagnostic condi-tion, a number of covariate analyses were per-formed. First, we considered the possible roleof length of time in the United States as a fac-tor. Because second-generation students havespent more time in the United States, it is like-ly that they have had greater exposure to itsnorms and culture, including the negativestereotypes about African Americans.However, a covariance analysis with time inthe United States had relatively little impacton the finding: the covariate itself was not sig-nificant, and its inclusion shifted the signifi-cance level only slightly to 0.054. We alsoexamined the correlation between time in theUnited States and the various measures of per-formance within the first generation only (andseparately for diagnostic and non-diagnosticconditions). No significant relationships werefound, suggesting that time in the UnitedStates, though a frequently-used demographicindex, is not a reliable gauge for psychologicalprocesses that may or may not occur in a giveninterval of time. Additionally, an examinationof gender differences showed that men andwomen performed equally in both diagnosticand non-diagnostic conditions. Thus the dif-ferences between the generations in the twoexperimental conditions appear quite robust.

Experimenter Effects

It was anticipated that race of experi-menter might affect performance, such thatstereotype threat effects would be more pro-

nounced when the experimenters were white.This possibility, which would be evidenced ina significant three-way interaction, was notsupported. Race of experimenter did have asignificant influence on performance, howev-er, as evidenced by significant two-way inter-actions between experimenter and diagnosticcondition ( p < .001), as well as betweenexperimenter and generation ( p < .05).

With regard to the first of these interac-tions, the pattern of means shows that perfor-mance in the diagnostic condition is enhancedwhen the experimenter is white (M = 46.2, SD= 17.7 for white experimenters, M = 34.2, SD= 16.7 with black experimenters). In contrast,performance in the non-diagnostic conditionis better when the experimenter is black (M =43.9, SD = 15.1 for black experimenter, M =34.7, SD = 10.2 for white experimenter).

Analysis of the significant interactionbetween generation and experimenter alsoshows reverse patterns for the black and whiteexperimenters. First-generation students dobetter when the experimenter is white (M =47.7, SD = 15.7 for white experimenter, M =38.3, SD = 14.7 for black experimenter).Second-generation students, in contrast, dobetter when the experimenters are black (M =44.2, SD = 15.6) than when they are white (M= 36.5, SD = 14.7). We consider the possiblemeaning of these somewhat unexpected pat-terns in the discussion.

Ethnic Identification and Stereotype Threat

Correlational analyses indicated that gen-eration and ethnic identification are related,both in the larger preliminary sample, r (262)= .37, p < .001, as well as in this experimentalsubsample, r (73) = .43, p < .001. This rela-tionship suggested two analytic strategies.First we used ethnic identification as a covari-ate in the generational analysis. This inclusiondid not substantially alter the obtained interac-tion between generation and diagnosticity(nudging the significance level from 0.036 to0.051).

Another way to test the influence of eth-nic identification was to use this variable as afactor in the analysis of variance design.Because ethnic identification and generation


4 A one-tailed test (based on Levene’s test assumingunequal variances) was used, consistent with our direc-tional prediction for the diagnostic condition.


covary, and because the sample size was limit-ed, it was not possible to factorially vary thetwo within a single analysis. Consequently, wechose to analyze the data in a second 2 � 2 �2 factorial design, this time substituting ethnicidentification for generation. As noted earlier,ethnic identification scores in the sample weregenerally skewed toward identification asWest Indian. Thus, in this analysis we dividedthe sample on the basis of stronger or weakerWest Indian identification, such that thosewho responded with a 1 or 2 (definitely ormostly West Indian) are contrasted with thosewho responded with either 3, 4, or 5 (definite-ly African American through an equalendorsement of the two ethnic labels).

Results of this analysis, focusing on thetwo-way interaction between ethnic identifica-tion and diagnostic condition, showed theanticipated pattern of means, although theinteraction itself did not attain conventionallevels of significance, F (1,65) = 2.50, p = .12.Thus, although the pattern that we find is con-sistent with the hypothesis that students whoidentified more with African American will bemore susceptible to the effects of stereotypethreat, the lack of a sharp demarcationbetween the two groups in terms of their eth-nic identification does not allow a strong testof the hypothesis.

Discussion

Differences between first- and second-generation West Indian immigrants, suggestedin more macrolevel data such as educationaland occupational achievement, clearly find aparallel here at a psychological level.Although the difference in the time that first-and second-generation students have spent inthe United States is only 11.5 years, their per-formance under conditions of stereotypethreat differs significantly. Both groups areaffected by the instructions, but in oppositedirections. Second-generation students show apattern of performance that is similar to thatreported previously for African American stu-dents (Steele and Aronson 1995; Massey andFischer 2005), as well as for other groups forwhom negative stereotypes about their group’sabilities exist (e.g., women in math, Spencer,

Steele, and Quinn 1999; people who have lowsocioeconomic status, Croizet and Claire1998). Specifically, when stereotype threat ispresent, their performance drops in compari-son to the first-generation comparison groupand to their own performance when the test isnon-diagnostic.

When conditions do not make stereotypethreat salient, first- and second-generationstudents perform equivalently. This equiva-lence of first- and second-generation studentsin the non-diagnostic condition is important,as it rules out alternative explanations basedon assumed differences between the twogroups in terms of competency or general test-taking ability. The difference between thegroups is not in their capability to perform onthe test (or their academic preparation andmotivation to perform), but in the specifics ofthe testing condition that bring other influ-ences to bear. Further, although not testablehere, the results also give us some reason toquestion the utility of those explanations ofdifferences between West Indians and AfricanAmericans that rely on dispositional factors,such as migration selectivity and differentialability or motivation.

First-generation immigrants show a quitedifferent reaction to stereotype threat condi-tions. Not only do they not show a decrement,but in fact their performance increases, rela-tive to the non-threat condition. The seemingimmunity of first-generation West Indianimmigrants to stereotype threat effects cannotbe explained either by a lack of awareness ofthe prevailing images of African Americans,nor by the absence of experience with race-based discrimination. Indeed, despite their rel-atively short time in the United States, first-generation immigrants do not differ in eitherrespect from their second-generation counter-parts who were born in the United States. Bothare equally likely to expect discrimination andto anticipate anxiety in the variety of socialsituations tapped by the measure of Sensitivityto Race-based Discrimination. Both see a dif-ference in the images that society has ofAfrican Americans in contrast to West Indians.

The difference between the two genera-tions may be explained in part by the positivi-ty of metastereotype of West Indians. First-



generation students are more positive in theirappraisal, believing that others view WestIndians more favorably than do the second-generation students (evidenced both in mea-sures of stereotypes and in the public regardsubscale of the Collective Self-esteem Scale).Accordingly, first-generation students appearable to turn to a positive image of their groupin the face of diagnostic pressure and to dis-tance themselves more effectively from thenegative stereotypes associated with blackperformance in the United States. Recentwork by Gilkes (2005), who interviewed first-generation West Indian immigrants in NewYork City and Toronto, is also consistent withthis analysis, showing that a positive view ofone’s group serves as a source of resilience inthe context of discriminatory treatment. Thisdistancing strategy appears to have limitedutility for the second generation, however,who are more aware of and immersed in theprejudicial views against blacks that persist inthe United States.

The first-generation pattern of increasedperformance in the face of possible threat isalso suggestive of the stereotype lift effect thatWalton and Cohen (2003) have documented.As defined by Walton and Cohen, “stereotypelift is the performance boost caused by theawareness that an outgroup is negativelystereotyped” (2003:456). In a review of previ-ous research, they show that people who aremembers of non-stereotyped groups (e.g.,men, whites) are also aware of the stereotypes.When put in a situation that brings thosestereotyped groups to mind, people in themore favorable group engage in a process ofdownward comparison that in turn elevatestheir sense of self-efficacy. The consequenceis a significant increase in performance, ascompared to a non-threat control condition.Further, Walton and Cohen (2003) concludethat stereotype lift effects are particularly like-ly when people believe in the validity of thenegative stereotypes about the other group orwhen they endorse the legitimacy of a group-based hierarchy. It is possible that some or allof these conditions are true for first-genera-tion immigrants. As we have shown, they areaware of the African American stereotype, andthere is literature to suggest that they see dif-

ferences between themselves and AfricanAmericans (Vickerman 1999; Waters 1999a).Experimental work by Barreto, Ellemers, andPalacios (2004) also points to the relevance ofone’s past experience with discrimination.They find that people who have a collectivehistory of success will view situations inwhich they are a token as a challenge and per-form better. In contrast, those who are histori-cally disadvantaged are more likely to experi-ence token situations as threatening and toperform worse. By extension, first-generationimmigrants who have lived in a society inwhich their group was the ruling majority maysee the diagnostic situation as a challenge, incontrast to second-generation immigrantswhose historical basis is the race-based statushierarchy of the United States.

Ethnic Identification

Generational effects in performance areparalleled by differences in ethnic identifica-tion. Second-generation West Indian immi-grants are significantly less likely to identifyas West Indian. It is important to note, howev-er, that even second-generation students in thissample were more identified as West Indianthan as African American, as indicated bytheir mean scores on the bipolar scale. Thisskew in the sample makes a strong test of theethnic identification and stereotype threathypothesis difficult, in that those who mightbe most strongly identified as AfricanAmerican are not well represented in the sam-ple. Nonetheless, the fact that we found a cleartrend in the predicted direction, in the face ofrelatively weak differences in ethnic identifi-cation, suggests that the influence of identifi-cation is a real phenomenon.

It is interesting to compare the patterns ofethnic identification found here with the dataof Waters (1994), obtained from adolescentsin a similar locality. Waters found roughlyequal proportions of students claimingAfrican American versus country of originidentification. In contrast, only 12% of ourstudents identified as either wholly or mostlyAfrican American. One source of the differ-ence may be related to the demographic pat-terns that Waters found, wherein adolescents



who were more identified as AfricanAmerican were also more likely to live inpoorer neighborhoods and to attend lowerquality schools. Thus, it is probable that stu-dents in her sample who identified as AfricanAmerican are less likely to attend college;consequently, a sample of college students(even in this case at a large public university)are likely to show a skewed distributiontoward the West Indian end of the identifica-tion dimension.5

This study was done with undergraduatestudents at a public university in New YorkCity, a context whose particular features needto be kept in mind. Accordingly, we can notgeneralize to the population of West Indianimmigrants at large. On the one hand, theresults of Massey and Fischer (2005) showthat stereotype threat effects are evident in stu-dents of color across a broad range of moreelite universities (although less pronouncedwhen representation of minority faculty wasgreater). At the same time, the general compo-sition both of the immediate university settingand the larger community could well be influ-ential. Stereotype lift effects might be morecommon, for example, when there is a viableWest Indian community with which one canidentify.

Although we talk in a language of con-trasts convenient to experimental design, thatis, a contrast between identification as WestIndian or African American, it is evident to usthat ethnic identification in immigrant com-munities is resistant to such simpledichotomies. Open-ended responses to a ques-tion asking for ethnic identification frequentlyelicit multiple terms that include references toboth race and ethnicity. Both the labels them-selves and the meanings associated with thecategories differ among first- and second-gen-eration Afro Caribbean immigrants(Butterfield 2004; Foner 2005; Gilkes 2005).Butterfield (2004), for example, argues thatsecond-generation West Indian immigrantstypically have both a racial and an ethnic iden-

tity, distinct in form but equally important toself-definition. This multidimensionality ofethnic identification suggests both caution andchallenge when conducting research withimmigrant populations.

Experimenter Effects

We had anticipated a possible three-wayinteraction when race of experimenter wasincluded in the analysis, such that the predict-ed two-way interaction between generationand diagnostic condition would be more like-ly to occur with white experimenters and lesslikely, or even absent, in the presence of blackexperimenters. This three-way interaction didnot emerge. However, race of experimenterclearly influenced the performance of our par-ticipants, as shown in the two significant two-way interactions between experimenter andboth diagnostic condition and generation.White experimenters elicited better perfor-mance when the test was diagnostic and whenthe participants were first-generation; blackexperimenters elicited better performancewhen the test was non-diagnostic and whenthe participants were second-generation.

The finding that second-generation stu-dents do better with a black experimenter isconsistent with the Massey and Fischer (2005)findings that a context with a higher percent-age of minority faculty was more favorable forblack and Latino students. The differencebetween first- and second-generation studentsin these different experimenter conditions isperhaps more easily understood if we consid-er the differing metastereotypes of first- andsecond-generation students. First-generationimmigrants believe that their group is regard-ed more favorably by society in general, whichwe can assume would be defined largely interms of the white majority. Accordingly, awhite experimenter may act as a proxy for thatlarger reference group and lead participants towant to live up to that perceived standard anddistinguish themselves from the AfricanAmerican image that they view as significant-ly less favorable. Second-generation students,in contrast, have a diminished view of themetastereotype of West Indians and are at thesame time more likely to see themselves as


5 We note that ethnic identification as an AfricanAmerican or West Indian was not correlated with stu-dents’ perceptions of class status within our larger sample(r = .11).


defined by the stereotype of AfricanAmericans. This shifting identification maylead them to be more comfortable with a blackexperimenter and less comfortable with awhite experimenter whom they might assumeis negatively disposed toward them.

For first-generation students, the whiteexperimenter was associated with enhancedperformance. To the extent that first-genera-tion students believe that the society at large(which is predominantly white) has a favor-able view of West Indians, the white experi-menter may serve as a cue for enhanced per-formance. Indeed, the stereotype lift effectsuggested earlier may be enhanced by thepresence of a white experimenter, who is moreclosely associated with the dominant andjudging society.

Future Directions

A key question for future research con-cerns the possible mediators of the generation-performance relationship. Although identifi-cation is clearly related to generation, as wellas to stereotype threat, it alone can not accountfor the differences between first- and second-generation respondents. As is the case forstereotype threat research in general, thesearch for mediating mechanisms continues.One possibility is that differences in motiva-tional states or self-regulatory focus (seeHiggins 1997) may be related to the genera-tional patterns. Keller and Blass (2003) havereported that the manipulation of a preventionfocus facilitates stereotype threat effects,whereas manipulating a promotion focus min-imizes such effects. Following this logic, itmight be that first-generation West Indians aremore characterized by a promotion focus, inwhich the emphasis of achievement is on theapproach to a desired goal. Second-generationWest Indians, in contrast, because of moreexperiences with race-based discrimination inthe United States, might be more motivated bya prevention strategy, in which achievement isseen as the avoidance of possible negativeevents.

The generality of these findings for otherimmigrant populations is also of considerableinterest. Immigrants from Mexico, for exam-

ple, are often burdened with negative stereo-types of intellectual capability as well, andcould be subject to the same kind of stereo-type threat effects as we observed here amongWest Indians. In the case of Mexicans, how-ever, it is not certain whether first-generationimmigrants would be impervious to stereotypethreat effects, given the negative stereotypesthat often characterize both Mexican nationalsand Mexican immigrants. Thus, first-genera-tion Mexican immigrants might show equal oreven greater stereotype threat effects thanwould later generations. In contrast, we mightthink about a group such as Asian immigrants,for whom the stereotype of a high-performing“model minority” is frequently invoked.Would stereotype lift effects be more evidentin this group when the group stereotype wasmade salient? These questions speak to thecomplexity of the immigrant experience andthe need for careful consideration of specificstereotypes and contexts.

The findings reported here contribute tothe burgeoning literature on stereotype threateffects, adding immigration status to the cate-gories of persons potentially subject to perfor-mance decrements under threat conditions.More importantly, however, the results speakto the dynamic relationship between personand context. West Indian immigrants do notautomatically become subject to stereotypethreat because they are black; rather they learnto experience stereotype threat as a result oftheir socialization into U.S. society wherebeing categorized as black has negative con-tingencies. Because these connections arelearned, one has to allow the possibility that,given optimal interventions, they can be over-come as well.

These results also speak to the importanceof understanding social psychologicalprocesses inherent to the immigration experi-ence. Sociologists and demographers havegenerated a wealth of data on the fortunes andadaptations of immigrants, including occupa-tional status, educational outcomes, and inter-marriage patterns. Underlying these group-level analyses are a wealth of psychologicalprocesses that need to be unpacked and artic-ulated. We suggest here that stereotype threatis one of the psychological processes that may



be at work. We of course recognize that ourdata are based exclusively on college students,considering the ways in which a form of acad-emic performance can be affected by stereo-type threat. Whether this process could also bedetected in more complex settings remains tobe established (though the work of Masseyand his colleagues in a broader range of edu-cational settings attests to the likely generaliz-ability of the phenomena).

More than simply an account of the num-bers of people entering a country, immigrationis a dynamic process of interplay betweenpeople and their social and cultural environ-ment. Further analysis of the psychologicalprocesses will allow us to determine what fac-tors either enhance or impede the participationof immigrants in their new culture.

REFERENCES

Aronson, Joshua., Michael J. Lustina, Catherine Good,Kelli Keough, Claude M. Steele, and JosephBrown. 1999. “When White Men Can’t DoMath: Necessary and Sufficient Factors inStereotype Threat.” Journal of ExperimentalSocial Psychology 35:29–46.

Ashmore, Richard D., Kay Deaux, and TracyMcLaughlin-Volpe. 2004. “An OrganizingFramework for Collective Identity: Articulationand Significance of Multidimensionality.”Psychological Bulletin 130:80–114.

Barreto, Manuela, Naomi Ellemers, and MariaSoledad Palacios. 2004. “The Backlash of TokenMobility: The Impact of Past Group Ex-periences on Individual Ambition and Effort.”Personality and Social Psychology Bulletin30:1433–1445.

Butterfield, Sherri-Ann, P. 2004. “’We’re Just Black’:The Racial and Ethnic Identities of Second-gen-eration West Indians in New York.” Pp. 288–312in Becoming New Yorkers: Ethnographies of theNew Second Generation, edited by PhilipKasinitz, John H. Mollenkopf, and Mary C.Waters. New York: Russell Sage Foundation.

Croizet, Jean-Claude and Theresa Claire. 1998.“Extending the Concept of Stereotype Threat toSocial Class: The Intellectual Under-performance of Students from Low Socio-economic Backgrounds.” Personality and SocialPsychology Bulletin 24:588–594.

Danso, Henry A. and Victoria M. Esses. 2001. “BlackExperimenters and the Intellectual TestPerformance of White Participants: The Tablesare Turned.” Journal of Experimental SocialPsychology 37:158–165.

Deaux, Kay. 1996. “Social Identification.” Pp.

777–798 in Social Psychology: Handbook ofBasic Principles, edited by E. Tory Higgins andArie W. Kruglanski. New York: Guilford Press.

———. 2006. To Be an Immigrant. New York:Russell Sage.

Foner, Nancy. 2001. Islands in the City: West IndianMigration to New York. Berkeley, CA:University of California Press.

———. 2005. In a New Land: A Comparative View ofImmigration. New York: New York UniversityPress.

Gilkes, Alwyn D. 2005. Among Thistles and Thorns:West Indian Diaspora Immigrants in New YorkCity and Toronto. PhD dissertation, CityUniversity of New York Graduate Center.

Higgins, E. Tory. 1997. “Beyond Pleasure and Pain.”American Psychologist 52:1280–1300.

Kalmijn, Matthijs. 1996. “The SocioeconomicAssimilation of Caribbean American Blacks.”Social Forces 74:919.

Kasinitz, Philip and Jan Rosenberg. 1996. “Missingthe Connection: Social Isolation andEmployment on the Brooklyn Waterfront.”Social Problems 43:180–197.

Katz, Irwin and Charles Greenbaum. 1963. “Effects ofAnxiety, Threat, and Racial Environment onTask Performance on Negro College Students.”Journal of Abnormal and Social Psychology66:562–567.

Katz, Irwin, S. Oliver Roberts, and James M.Robinson. 1965. “Effects of Task Difficulty,Race of Administrator, and Instructions onDigit-symbol Performance of Negroes.” Journalof Personality and Social Psychology 2:53–59.

Keller, Johann and H. Bless. 2003. “Regulatory Focusas a Moderator of Stereotype Threat Effects.”Paper presented at meeting of the Society forPersonality and Social Psychology, LosAngeles, CA.

Logan, John R. and G. Deane. 2003. Black Diversity inMetropolitan America. Albany, NY: LewisMumford Center for Comparative Urban andRegional Research, University of Albany (http://web.archive.org/web/20040317091934/mumford1.dyndns.org/cen2000/BlackWhite/BlackDiversityReport/Black_Diversity_final.pdf)

Luhtanen, Riia and Jennifer Crocker. 1992. “ACollective Self-esteem Scale: Self-evaluation ofOne’s Social Identity.” Personality and SocialPsychology Bulletin 18:302–318.

Massey, Douglas S., Camille Z. Charles, Garvey F.Lundy, and Mary J. Fischer. 2003. The Source ofthe River. Princeton, NJ: Princeton UniversityPress.

Massey, Douglas S. and Mary J. Fischer. 2005.“Stereotype Threat and Academic Performance:New Findings from a Racially Diverse Sampleof College Freshmen.” Du Bois Review 2:45–67.

Mendoza-Denton, Rodolfo, Geraldine Downey,



Kay Deaux is Distinguished Professor of Psychology at the Graduate Center of the City Universityof New York. Together with colleagues and students, her work focuses on the social psychologicalaspects of immigration, as conceptualized in her recent book, To Be an Immigrant (2006). She is the2007 recipient of the Kurt Lewin Memorial Award from the Society for the Psychological Study ofSocial Issues.

Nida Bikmen received her PhD in Social/Personality Psychology from the Graduate Center of theCity University of New York. She is currently a visiting assistant professor in the PsychologyDepartment at Denison University in Granville, Ohio. She is interested in how ethnic and nationalidentities shape and are shaped by interethnic conflict and international migration.


Valerie Purdie, Angelina Davis, and JaninaPietrzak. 2002. “Sensitivity to Status-basedRejection: Implications for African AmericanStudents’ College Experience.” Journal ofPersonality and Social Psychology 83:896–918.

Model, Suzanne. 1991. “Caribbean Immigrants: ABlack Success Story?” International MigrationReview 25:248–257.

———. 1995. “West Indian Prosperity: Fact orFiction.” Social Problems 42:535–544.

Phinney, Jean S. 1998. Multi-ethnic identity measure(revised version). Personal communication.

Portes, Alejandro. 1995. The Economic Sociology ofImmigration: Essays on Networks, Ethnicity andEntrepreneurship. New York: Russell SageFoundation.

Portes, Alejandro and Rubén Rumbaut. 1992.Immigrant America. Berkeley: University ofCalifornia Press.

———. 2001. Legacies: The Story of the ImmigrantSecond Generation. Berkeley, CA: University ofCalifornia Press.

Reid, Ira D. A. 1939. The Negro Immigrant: HisBackground, Characteristics and Social Adjust-ment, 1899–1937. New York: ColumbiaUniversity Press.

Schmader, Toni. 2002. “Gender IdentificationModerates Stereotype Threat Effects onWomen’s Math Performance.” Journal ofExperimental Social Psychology 38:194–201.

Sigelman, Lee, and Steven A. Tuch. 1997.“Metastereotypes: Blacks’ Perceptions ofWhites’ Stereotypes of Blacks.” Public OpinionQuarterly 61:87–101.

Spencer, Steven J., Claude M. Steele, and Diane M.Quinn. 1999. “Stereotype Threat and Women’sMath Performance.” Journal of ExperimentalSocial Psychology 35:4–28.

Steele, Claude M. 1997. “Threat in the Air: HowStereotypes Shape Intellectual Identity andPerformance.” American Psychologist 52:613–628.

Steele, Claude M., and Joshua Aronson. 1995.“Stereotype Threat and the Intellectual TestPerformance of African Americans.” Journal ofPersonality and Social Psychology 69:797–811.

Steele, Claude M., Steven J. Spencer, and Joshua

Aronson. 2002. “Contending with Group Image:The Psychology of Stereotype and SocialIdentity Threat.” Advances in ExperimentalSocial Psychology, 34:379–440.

Tormala, Teceta Thomas. 2005. “The Meaning ofBlackness: Stereotypes about Immigrant andAmerican Blacks.” Manuscript in preparation.

Tilly, Charles. 1990. “Transplanted Networks.” Pp.79–95 in Immigration Reconsidered: History,Sociology and Politics, edited by Virginia Yans-McLaughlin. New York: Oxford UniversityPress.

Torres, Kim C. and Camille Z. Charles. 2004.“Metastereotypes and the Black-White Divide:A Qualitative View of Race on an Elite CollegeCampus.” Du Bois Review 1:115–149.

Vickerman, Milton. 1999. Crosscurrents: West IndianImmigrants and Race. New York: OxfordUniversity Press.

Walton, Gregory M. and Geoffrey L. Cohen. 2003.“Stereotype Lift.” Journal of ExperimentalSocial Psychology 39:456–457.

Waters, Mary C. 1990. Ethnic Options. Berkeley, CA:University of California Press.

———. 1994. “Ethnic and Racial Identities ofSecond-generation Black Immigrants in NewYork City.” International Migration Review28:795–820.

———. 1999a. Black Identities: West IndianImmigrant Dreams and American Realities.New York: Russell Sage Foundation.

———. 1999b. “West Indians and AfricanAmericans at Work: Structural Differences andCultural Stereotypes.” Pp. 194–227 inSociodemographic Aspects of Immigration forAfrican Americans, edited by Frank D. Bean andStephanie B. Rose. New York: Russell SageFoundation.

Waters, Mary, Philip Kasinitz, John Mollenkopf,Nancy Lopez, and Dae Young Kim. 1998. “TheSchool to Work Transition of Second GenerationImmigrants in Metropolitan New York: SomePreliminary Findings.” Paper prepared forWeiner Inequality and Social Policy SeminarSeries, Harvard University, Cambridge, Mass.,November 23.


Alwyn Gilkes received his PhD in Social/Personality Psychology from the Graduate Center of theCity University of New York and is currently Adjunct Professor in the Psychology Department atMedgar Evers College of CUNY. He is the author of The West Indian Diaspora: Experiences in theUnited States and Canada (2007). His research interests include immigrant acculturation, education-al success and achievement, and resilience.

Ana Ventuneac is a PhD candidate in Social/Personality Psychology at the Graduate Center of theCity University of New York, completing her dissertation on the role of social stress in sexual riskbehaviors. She is Project Manager at the HIV Center for Clinical and Behavioral Studies at New YorkState Psychiatric Hospital.

Yvanne Joseph is a Lecturer and Counselor in the SEEK Program at Medgar Evers College of theCity University of New York and an advanced doctoral student in the Social/Personality Psychologyprogram at the CUNY Graduate Center. Her research concerns the social and subjective experienceof being black, immigrant, and bilingual in the United States.

Yasser A. Payne is Assistant Professor in the Black American Studies Program at the University ofDelaware. He received his PhD in Social/Personality Psychology from the Graduate Center of theCity University of New York and had a National Institute of Drug Abuse (NIH-NIDA) postdoctoral fel-lowship. In his research, he uses participatory action methods to study resiliency with street-life ori-ented black men. He is coauthor of Echoes of Brown: Youth Documenting and Performing the Legacyof Brown v. Board of Education (2004).

Claude M. Steele is the Director of the Center for Advanced Study in the Behavioral Sciences inStanford, California and the Lucie Stern Professor in the Social Sciences at Stanford University.Throughout his career, he has been interested in how people cope with threats to their self-image; hisdevelopment of the concept of stereotype threat has been particularly important in illuminating theways in which societal stereotypes have an impact on individual performance.


becoming american: stereotype threat effects in afro ... · afro-caribbean would include people...

Documents