Article Critique Documents Attached

in as

ses e ss st

Journal of Experimental Social Psychology38, 113–125 (2002) doi:10.1006/jesp.2001.1491, available online at http://www.idealibrary.com on

Reducing the Effects of Stereotype Threat on African American College Students by Shaping Theories of Intelligence

Joshua Aronson

New York University

Carrie B. Fried

Winona State University

and

Catherine Good

The University of Texas at Austin

Received March 23, 2000; revised May 10, 2001; accepted May 10, 2001; published online July 18, 2001

African American college students tend to obtain lower grades than their White counterparts, even when they enter college with equivalent test scores. Past research suggests that negative stereotypes impugning Black students’ intellectual abilities play a role this underperformance. Awareness of these stereotypes can psychologically threaten African Americans, a phenomenon known “stereotype threat” (Steele & Aronson, 1995), which can in turn provoke responses that impair both academic performance and psychological engagement with academics. An experiment was performed to test a method of helping students resist these respon to stereotype threat. Specifically, students in the experimental condition of the experiment were encouraged to see intelligence—th object of the stereotype—as a malleable rather than fixed capacity. This mind-set was predicted to make students’ performances le vulnerable to stereotype threat and help them maintain their psychological engagement with academics, both of which could help boo

their college grades. Results were consistent with predictions. The African American students (and, to some degree, the White students) encouraged to view intelligence as malleable reported greater enjoyment of the academic process, greater academic engagement, and obtained higher grade point averages than their counterparts in two control groups.© 2001 Elsevier Science (USA)

an etal

ality end

blem cks

ich way

as t to eve- our- y to tend ri-

firs n and eline orris ulad n for

The traditional model [of intelligence] may be a cause of rather th a potential answer to educational problems, in particular, and soci problems, in general.

—Robert Sternberg (1998)

Because education is the surest route to social equ the academic underachievement of Black Americans t

This research was supported by an NSF CAREER grant to the author as well as by funding from the James S. McDonnell Foundatio The Spencer Foundation. We are grateful to Eron Al-Amin, Emm Chen, Cinahi Coleman, Angela Haydel, Matthew Mizel, and Edwin M for assistance with the conduct of the research and to Rachel Fo Matthew McGlone, Tor Neilands, Steven Spencer, and Timothy Wilso helpful comments on the manuscript.

Address correspondence and reprint requests to Joshua Aronson, D partment of Applied Psychology, New York University, New York, NY 10003. E-mail: [email protected].

113

, s

to be regarded as both an educational and a social pro (e.g., Garibaldi, 1991; Herrnstein & Murray, 1994; Jen & Phillips, 1998). It is a problem, we believe, to wh Sternberg’s argument is particularly applicable; the people traditionally have thought about intelligence— largely unmodifiable—is more a barrier than a boos African American achievement and indeed, the achi ment of all students (e.g., Schwartz, 1997). Could enc aging a view of intelligence as expandable hold a ke educational improvement for Black students, who con regularly with debilitating suspicions of intellectual infe ority? Our research explores this question.

AFRICAN AMERICAN UNDERACHIEVEMENT

t

i,

There is much disagreement about the underlying causes for and, hence, the wisest remedies to the problem of

e-

0022-1031/01 $35.00 © 2001 Elsevier Science (USA)

All rights reserved.

uite re o oling rts. e a son ele, pro

gard an f th

fierc ocus neti tein de do an ong are

ans less re o

netic arly hus gies

S

to a ear eri- bou as

n et d is

ed i as

eo- typ an k in

e an reo f a the

or-

t the dent n, & ereo-

ain e the

al.,

emic un, ous ge son, eele siz-

bility race for- s the owed har- ity— om- the

ies, s or

igher

rmine ogi-

help ance

at to ntify s in

rter, eed in n if rede- in no nson, ele,

cess itua- ative one tage

le, a e to d of t of

ime, %—

RIED

African American educational achievement. But q clearly, a student’s race matters. Virtually every measu academic achievement taken at every level of scho shows African Americans trailing their White counterpa This gap in test and academic performance cannot b tributed entirely to socioeconomic status (see Aron Quinn, & Spencer, 1998; Jencks & Phillips, 1998; Ste 1997, for extensive discussions of race gaps and their posed causes). Although there is little disagreement re ing the scope or gravity of the problem of African Americ underachievement, attempts to pin down the causes o problem have fueled decades of visceral debate. Most are the well-known nature–nurture debates, which f upon whether performance differences stem from ge cally determined differences in intelligence (e.g., Herrns & Murray, 1994) or from environmental factors that impe certain groups from developing the skills they need to well on tests and in school (see Jacoby & Glauberm 1995, for a review). But there is also disagreement am environmentalists regarding which structural factors most to blame for the poor outcomes for African Americ and other ethnic minorities (Neisser, 1986). Neverthe the most widely cited causes, be they endowed by natu imposed by society, are largely intractable. That is, ge predisposition, poverty, culture, and the like are cle factors that are difficult, if not impossible, to alter, and t such explanations offer little in the way of specific strate for addressing race gaps in performance.

STEREOTYPE THREAT AND ITS ROLE IN RACE GAP

In contrast, a good deal of recent research points psychological factor in this underachievement that app to be far more amenable to intervention—African Am cans’ responses to stereotypes alleging inferior ability a their group. This psychological factor is referred to “stereotype threat” (e.g., Aronson et al., 1999; Aronso al., 1998; Steele, 1997; Steele & Aronson, 1995), an described as a social psychological predicament root the prevailing American image of African Americans intellectually inferior.1 The basic notion behind the ster type threat analysis is this: in situations where a stereo about a group’s intellectual abilities is relevant—taking intellectually challenging test, being called upon to spea class, and so on—Black students bear an extra cognitiv emotional burden not borne by people for whom the ste type does not apply. This burden takes the form o performance-disruptive apprehension, anxiety about possibility of confirming a deeply negative racial inferi

1 Polls indicate that a sizable portion of White Americans—53

114 ARONSON, F

indicate thinking Blacks to be less intelligent than Whites (Smith, 1990), and even greater numbers appear to implicitly hold this stereotype (Devine 1989).

f

t- ,

- -

e e

-

,

, r

s

t

n

e

d -

ity—in the eyes of others, in one’s own eyes, or both a same time. Importantly, it is not necessary that a stu believe the stereotype to feel this burden (Good, Aronso Harder, 2000). He or she need only be aware of the st type and care enough about performing well in the dom (e.g., on the test, in the math class) to want to disprov stereotype’s unflattering implications (e.g., Aronson et 1999; Aronson & Good, 2000).

Stereotype threat appears to undermine acad achievement primarily in two ways. First, in the short r it can impair performance by inducing anxiety. Numer laboratory experiments involving African-American colle students have documented this short-term effect (Aron 1999; Aronson et al., 1998; Blascovich et al., 2001; St & Aronson, 1995). Inducing stereotype threat— empha ing intelligence by presenting a test as a measure of a or emphasizing race by having test-takers indicate their on the test booklet—significantly undermined the per mance of African Americans on intellectual tests such a Graduate Record Exam (GRE). The same studies sh how minimizing stereotype threat—for example, by c acterizing a standardized test as nondiagnostic of abil significantly improved performance, in many cases c pletely eliminating the performance gaps between African Americans and Whites. In many of these stud levels of anxiety (as measured by self-report inventorie direct measures of blood pressure) were significantly h under conditions of stereotype threat.

The second way stereotype threat appears to unde achievement is through “disidentification,” the psychol cal disengagement from achievement hypothesized to students cope with stereotype threat and underperform in a given domain. Many researchers have noted th promote and maintain self-esteem, students tend to ide with—that is, to base their self-esteem upon— domain which they can excel (e.g., Eccles & Wigfield, 1995; Ha 1990). To sustain self-esteem one needs either to succ a domain—if one can— or to disidentify from the domai success is elusive. Disidentification occurs when one fines the self-concept such that a threatened doma longer is used as a basis of self-esteem (e.g., Aro Blanton, & Cooper, 1995; Pelham & Swann, 1989; Ste 1992). It should be noted that the disidentification pro can take various forms, ranging from temporary or s tion-specific devaluing of a domain in response to neg outcomes to a more chronic divestment of the self from or more domains of achievement. Devaluing, an early s of disidentification, can be observed when, for examp student proclaims that “math is for nerds,” in respons receiving a poor grade in math class. But often, this kin devaluing is short lived, a temporary disengagemen self-esteem from outcomes in a domain. But over t

, AND GOOD

chronic disengagement of this sort may lead the student to disidentify fully from mathematics (see Major & Schmader, ,

re i hrea un- &

ress o b et of an b , bot anc

ents

clea vert 995 -boo ator , two itua- e—

ican nson d be tone f rac om

l lab tive reo

ains be Af-

wn (e.g dren listi otyp ulate r pe gge ated bou

per- ll be intel this

a- pro- bility ruth n be ticity

or very - eve-

her ett,

ows lli- ich ry” they o in

to eg- their are ncre- ls.” epts chal- and rience en- an,

als rfor-

ty or rist n a redic- lec- he

to ereo- on- anip- ark

ental eo- -assur- r if od, nxiety vich

D TH

1998, for a thorough discussion of these issues). The increasing evidence that in part because of stereotype t African Americans are more prone than their White co terparts to disidentify from academics (e.g., Major Schmader, 1998; Osbourne, 1995; Steele et al., in p Because identification with academics is assumed t crucial for success in college or school, any force or s forces that frustrates this psychological engagement c a serious barrier to achievement (Steele, 1997). In sum responses to stereotype threat—impaired test perform and reduced identification— can critically depress stud performance in college.

As noted, the stereotype threat analysis offers the advantage of considering factors easier to change than po genes, and so forth (Steele, 1997; Steele & Aronson, 1 But easier does not necessarily mean easy. Performance ing factors that can be manipulated with ease in the labor may stubbornly resist change in schools. For example strategies— eliminating evaluative scrutiny in a testing s tion and making the test-takers’ racial identity a nonissu markedly improved the test performance of African Amer college students in laboratory studies (e.g., Steele & Aro 1995). But, in the typical college milieu, such steps woul next to impossible to take. Ability evaluation is a corners of schooling and the very essence of testing; awareness o and ethnicity is an inevitable feature of integrated classro or college campuses.

In the present research, rather than apply successfu oratory manipulations verbatim, we tried the alterna approach of using our understanding of the roots of ste type threat in the hopes of developing a protection ag some of its effects. Although an obvious tactic might simply to combat the prevailing stereotypes regarding rican-Americans’ intellectual abilities, such well-kno cultural stereotypes are notoriously resistant to change Hewstone, 1996; Pettigrew, 1981), even in young chil (e.g., Bigler, 1999). Thus, we reasoned that a more rea strategy would acknowledge the presence of the stere (and thus of stereotype threat), but attempt to inoc students against some of its undesired effects on thei formance and academic engagement. Past research su that such negative effects might be meaningfully attenu by encouraging students to change the way they think a intelligence itself. Specifically, we propose that under formance and devaluing of academic achievement wi lessened if stereotype targets are encouraged to see gence as modifiable. We now turn to the rationale for hypothesis.

CONCEPTIONS OF INTELLIGENCE AND STEREOTYPE THREAT

STEREOTYPE THREAT AN

Scholars have long argued about what intelligence is and how it should be measured (e.g., Gardner, 1983; Gould

s t,

). e

e h e ’

r y, ). st- y

,

e s

-

- t

.,

c e

r- sts

t

li-

1981; Herrnstein & Murray, 1994; Lewontin, Rose, & K min, 1984; Neisser et al., 1996). A question that has duced opinions at both extremes concerns the mallea of intelligence—whether it is expandable or fixed. The t appears to lie somewhere in between; intelligence ca expanded to some degree, but there are limits to its plas (e.g., Sternberg, 1996a). But irrespective of the truth— what psychometricians believe to be the truth—there is compelling evidence that what astudentthinks about intel ligence can have a powerful effect on his or her achi ment. The clearest example is provided by Dweck and colleagues (e.g., Dweck, 1999, 1986; Dweck & Legg 1988; Hong, Chiu, & Dweck, 1995). Their research sh how children’s implicit theories about the nature of inte gence—whether they tend to hold an “entity theory” (wh views intelligence as fixed) or an “incremental theo (which views it as malleable)— determines the goals pursue, their responses to difficulty, and how well they d school.

Students who hold an entity view of intelligence tend pursue “performance goals” (Dweck, 1999; Dweck & L gett, 1988); they are concerned with demonstrating intelligence and prefer tasks that will verify that they smart and capable. In contrast, students who hold an i mental view of intelligence tend to pursue “learning goa They tend to be more concerned with learning new conc and improving their competence. When tasks become lenging, entity theorists tend to become debilitated disengage, whereas malleable theorists appear to expe less anxiety, put forth more effort, and increase their gagement (Dweck, 1986, 1999; Nicholls, 1984; Utm 1997).

Our reasoning is that entity theorists and individu targeted by ability stereotypes may adopt the same pe mance goal mind-set when faced with academic difficul the possibility of low performance. Like the entity theo when faced with a difficult task, the Black student i stereotype threat situation faces essentially the same p ament, the troubling implication that he or she is intel tually limited, with little or no hope for improvement. T goal that flows from this mind-set is to protect—and project—an image of competence, to disprove the st type of intellectual inferiority. Consistent with this reas ing, past research has shown that stereotype threat m ulations elicit from stereotype targets many of the hallm responses that distinguish entity theorists from increm theorists (e.g., Dweck, 1999). Specifically, like entity th rists, stereotype targets tend to choose easier, success ing tasks when their abilities are subject to scrutiny o their ethnicity or gender is made salient (Aronson & Go 1999), experience greater performance pressure and a when tasks are both evaluative and challenging (Blasco

115EORIES OF INTELLIGENCE

, et al., 2001; Steele & Aronson, 1995), and tend to devalue ability domains in which they have performed poorly (e.g.,

we rt o mo- and or- con- -set h t

el- ay t in nce

heir

son- ma-

on mer- rba ility wer as e to

orted ity” tem udy val- pre- lity o ogus stu

este ace ow not nd- or

hus nde t th thre ans

our- nts e th

ht to thei nd i

s, the per- may avior king t the uilt not

evere

self- ge is

cular mes es, hown ., Pal- wn to vali-

Ross, e to with ause

tically , less ore d or ons ling influ- nce

g re- hus rrent

ude

both tudy. pro- iques

ake nd- and d to the

tion not . In

RIED

Major & Schmader, 1998; Major et al., 1998). In sum, suspect that negative ability stereotypes may derive pa their power to undermine intellectual performance and tivation precisely because they imply a self-threatening inalterable deficiency—a fixed lack of intelligence. Imp tantly, just as situations often influence people to act trary to their attitudes or personality traits, the mind imposed by stereotype threat may be strong enoug overwhelm an individual’s own implicit beliefs about int ligence. If our logic is correct, then it follows that one w to help students resist responding to stereotype threa maladaptive fashion—that is, by adopting a performa goal orientation—would be to convince them that t abilities are expandable.

Two recent laboratory studies support this line of rea ing. In the first study (Aronson, 1999), the effects of nipulating a fixed-versus-expandable view of an ability test anxiety and performance was examined. African A ican and White college students took a challenging ve test. Prior to the test, some were informed that the ab being tested was highly expandable, whereas others told that the ability was fixed. A third (control) group w simply told that the test measured verbal ability. Relativ the controls, test-takers (both Blacks and Whites) rep more anxiety and solved fewer items in the “fixed abil condition and reported less anxiety and solved more i in the “expandable ability” condition. In a second st examining the effect of an incremental mind-set on de uing (Aronson, 1997), Blacks and Whites took a test sented as measuring either an endowed and fixed abi an expandable skill. Following the test, they received b positive or negative performance feedback. Later the dents were asked how much they cared about the t ability. The results were very clear. Regardless of r students in the “fixed ability” condition who received a l score devalued the ability—that is, they claimed it was a particularly important skill to have. Those in the “expa able skill” condition, however, valued the skill whether not they thought they had performed well on the test. T encouraging students to see ability as exapandable u mined the two pernicious responses to stereotype threa have been proposed as mediators between stereotype and the poor achievement outcomes of African Americ

THE PRESENT STUDY

In the study to be reported, we wanted to see if enc aging incrementalism could be used to affect stude actual academic engagement and achievement outsid laboratory. Our objective was straightforward. We soug persuade a group of students to adopt the view that basic intelligence was malleable, that they could expa

116 ARONSON, F

with work. But, as research on attitude change shows per suasive messages often fail to move people if the issues a

f

o

a

l

e

s

r

- d ,

, r- at at

.

’ e

r t

important. Moreover, even when persuasion succeed attitude change may be short lived. And, even when suasion works well and the new attitude endures, it still lack the necessary cognitive accessibility to guide beh (see Petty & Wegener, 1998, for a review). Thus in see to create lasting and influential attitude change abou nature of intelligence, we created an intervention b around a variety of social psychological tactics shown only to change attitudes, but also to make them pers and come easily to mind.

Research, particularly within the dissonance and perception theory traditions, suggests that attitude chan greatly fostered by getting people to advocate a parti position in their own words, a phenomenon someti called the “saying-is-believing effect” (Higgins & Rhol 1978). Public commitment to an advocacy has been s to increase acceptance of the position advocated (e.g lak et al., 1981). Once formed, attitudes have been sho persevere and remain resistant to change, if they are dated by the message recipient’s own experiences ( Lepper, & Hubbard, 1975). In particular, inducing peopl consider how their own past behaviors are consistent an attitude strengthens that attitude (Fazio, 1995). Bec strong attitudes are more accessible and more automa activated, they are more resistant to counterinformation influenced by momentarily salient information, and m persistent over time. Consequently, attitudes change created in this way are most likely to influence acti (Fazio & Williams, 1986). Past interventions marshal such attitude change tactics have shown promise for encing such important behaviors as reducing viole among children (Huesman et al., 1983) and increasin cycling among adults (Fried & Aronson, 1995). We t attempted to integrate each of these tactics into the cu intervention to induce an influencial change in attit about the malleability of intelligence.

Method

Overview

Three groups of African-American and Caucasian ( male and female) undergraduates participated in the s One group participated in an intervention (a pen pal gram) that employed numerous attitude change techn designed to teach them, help them internalize, and m cognitively available the notion that intelligence is expa able (malleable pen pal condition). The attitudes achievement outcomes for this group were compare those of two control groups, one that participated in same intervention with a different intelligence orienta (control pen pal condition) and a third group that did participate in the intervention (non pen pal condition)

, AND GOOD

- re this way, we were able to determine whether the positive outcomes we were predicting resulted from adopting the

rtici- able ame par- tory ang

tici- telli- ining e st rtici s of heir ns ition oth two pa

able an

otyp ter-

an atio tere

d to ere

y of thei ents earc en- and mly

au- pal

wo d by they ses- ents p of laine them

were tran- sted ly 10 rtici-

cted, rade-

in ere d as trol

enter psy- stic as to ation- nts. nger l col-

come tro- par- tter ooler com- role on- able

, and ritten

tters was

s he vor- rtici- that

- age In of

par- t re- ence. view r an ith

D TH

malleable intelligence orientation, rather than mere pa pation in the pen pal program. Participants in the malle pen pal condition and the control pen pal condition c into the laboratory on three occasions, purportedly to ticipate in the pen pal program. During these labora sessions, the participants participated in the attitude ch intervention. A few days after the intervention the par pants completed a measure of their beliefs about in gence as a check on the manipulation, but the rema measures were given several months subsequent to th of the intervention. These later measures included pa pants’ beliefs about the nature of intelligence, rating their enjoyment of and identification with academics, t grades, and items designed to assess their perceptio stereotype threat. Participants in the non pen pal cond did not participate in the intervention but did complete b sets of measures. We predicted that relative to the control conditions, the participants in the malleable pen condition would come to see intelligence as more malle and, as a result, report greater academic identification enjoyment, higher grades, and, perhaps, less stere threat. We further predicted that the benefits of this in vention would occur primarily for the African Americ students, whose academic performance and identific we presumed to be depressed by their reactions to s type threat.

Participants and Design

A total of 109 Stanford undergraduates were recruite take part in the study for pay. A number of participants w unable to continue past the initial session for a variet reasons—time constraints, discomfort about releasing official grade transcripts, repeatedly missed appointm or because they asked (and were invited) to join the res team— but attrition did not vary as a function of experim tal condition. In the final analysis, a total of 79 male female participants (42 Black, 37 White) were rando assigned to one of the six conditions of the study, a 23 3 design yielded by crossing race (African American or C casian) with treatment (malleable pen pal, control pen or non pen pal).

Procedure

At the beginning of winter quarter (mid-January) t thirds of the participants—the pen pals—were contacte phone and scheduled for their first session. At this time, were told that they would be participating in several sions involving long-distance mentoring of young stud and a final unrelated study examining the relationshi some psychology measures and grades. It was exp that although these studies were unrelated, we wished

STEREOTYPE THREAT AN

to participate in all of them to simplify the bookkeeping on the research grant that was funding the project. None of the

e

art -

of

l

d e

n o-

r , h

,

d

participants questioned this reasoning. Participants asked to sign forms releasing their grade and SAT scripts from the registrar. The entire intervention consi of three 1-h laboratory sessions, spaced approximate days apart, and was completed by late February. Pa pants in the non pen pal control condition were conta scheduled, and asked to fill out measures and sign g release forms near the end of February.

Participants in the two pen pal conditions were run groups of two to five. Whenever possible the groups w racially mixed. Each session was randomly designate either the malleable intelligence orientation or the con orientation. Both began the same way. The experim (White female) introduced herself as an educational chologist working with an organization called “Schola Pen Pals.” The purported role of Scholastic Pen Pals w set up one-time letter exchanges between young, educ ally “at risk” middle school students and college stude The purpose of the exchange was to give the you students encouragement, to show them that successfu lege students had once been like them, but had over their struggles to find eventual success. After a brief in duction to the program’s procedures and philosophy, ticipants were informed that they would answer one le from a seventh grader. In each case, the middle-sch was characterized as coming from an impoverished munity and could thus benefit from having an elder model. The true purpose of the letter writing was to c vince half of the pen pals themselves of the expand nature of intelligence.

Middle school student letters.To increase believability all letters received by the participants were handwritten sealed in envelopes. Some students received letters w by boys, others received letters from girls, but the le made no reference to race. The content of the letters otherwise the same; the child described some difficultie or she was having in school in addition to describing fa ite subjects and activities. After reading the letters, pa pants were given instructions for writing their replies varied as a function of condition.

Malleable pen pal orientation. Participants in this con dition were asked to write a reply that would encour their pen pals to work hard in spite of their difficulties. addition to whatever they wanted to offer in the way encouragement, participants were told that it would be ticularly helpful to incorporate a theme stressing wha search was revealing about the nature of human intellig They were asked to impress upon their pen pals the that intelligence is not a finite endowment, but rathe expandable capacity that grows—“like a muscle”—w mental work. They were further told:

117EORIES OF INTELLIGENCE

Because intelligence is malleable, humans are capable of learning and mastering new things at any time in their lives. This message is especially important to get across to young, struggling students. If

hat ir ce ol

ar- how and

ded eu- wer tua

- ging un- s, in ese

ny tual ting ng ne globa ny

to

brie ning

osed

he , we the ffect tel-

onal the ind ha

their f th ts in am- nts ically po & fter the the

n pa em

eech le to ts in the

n- out

bility e re- their

with experi- is de-

ents’ Al- ental AT leable n

, n e

ar- ain

, ses

, we y of lace nter- s of d to ques- con- of

you asic an-

e (1) n the

RIED

these students view intelligence as a fixed quantity, they may feel t they are incapable of learning if they encounter difficulty with the school work. If, however, students can be convinced that intelligen expands with hard work, they may be more likely to remain in scho and put effort into learning.

To reinforce the scientific validity of this message, p ticipants were shown a brief video clip that discussed the brain, and hence intelligence, is capable of growing making new connections throughout life. The clip inclu a vivid color animation of the brain developing new n rons, while a voiceover reported that brain researchers discovering how the brain grows in response to intellec challenge.

Control pen pal orientation. This orientation was de signed to offer the same experience—writing encoura letters to a younger student—that differed only in the derlying message about the nature of intelligence. Thu addition to the same information given all pen pals, th participants were told that:

Intelligence is not a single entity, but rather composed of ma different talents, and, as a result, every person has both intellec strengths and weaknesses. Therefore, it is a potentially devasta mistake to view intelligence as a single attribute; it may lead you students to give up entirely on education if they are struggling in o subject, because the students can see themselves as failures at a level. But if struggling students can be convinced that there are ma different types of intelligence, they may be more likely to continue learn in an attempt to find and develop areas of strength.

To bolster this message, participants were shown a video clip that discussed how psychologists were begin to look at intelligence not as a single unit but as comp of many different abilities.

Attitude change tactics. Although the speech and t film clip seemed quite persuasive on their own, as noted wanted to maximize the durability and influence of attitude change. To engage this saying-is-believing e we asked participants to advocate the malleability-of-in ligence position. To bolster commitment to and pers responsibility for their message, a Polaroid photo of participant was taken and clipped to their letter. To rem them of their advocacy and to suggest that their letters impact, all participants received a thank-you note from pen pal and his or her teacher at the next session o intervention. To maximize belief perseverance, studen both conditions were asked to build into their letters ex ples from their own life that illustrated their argume about intelligence. To make the message as chron available as possible, we used repetition (e.g., Caciop Petty, 1979; Cook & Wadsworth, 1972). Specifically, a writing one letter, participants were brought back to laboratory on 2 subsequent days. On the second day wrote another letter with the same message to a new pe On the third day they reworked their letters, turning th

118 ARONSON, F

into brief speeches, which were then audiotaped for use in future interventions with at risk children. During this third

e l

l

f

,

d

e

y l.

session, participants listened to their own audiotaped sp twice. Because of these tactics, it seems reasonab assume that by the end of the third session, participan both conditions of the intervention were well versed in theory of intelligence assigned to them.

Dependent measures.Several days after the interve tion, as a check on the manipulation, participants filled a two-item measure assessing their belief in the mallea of intelligence. Several weeks later, these beliefs wer assessed in a separate survey along with ratings of enjoyment of academics, their degree of identification academics, and measures designed to assess their ence of stereotype threat. Each of theses measures scribed in detail in the next section.

Results and Discussion

SAT Scores

Subsequent to the completion of the study, stud official SAT scores were obtained from the registrar. though subjects were randomly assigned to experim condition, an initial observation of the means of the S scores suggest that, by chance, the subjects in the mal pen pal condition (M 5 1203) hadlower SAT scores tha subjects in both the pen pal control condition (M 5 1322) and the non pen pal condition (M 5 1261). Furthermore Black participants (M 5 1185) hadlower SAT scores tha White participants (M 5 1342). To determine if thes differences were significant, a 2 (race)3 3 (condition) analysis of variance (ANOVA) was performed on the p ticipants’ SAT scores. Results revealed a significant m effect of race, F(1, 73) 5 59.19, p , .001 and a significant main effect of condition,F(2, 73) 5 10.17 p , .001. To correct for these differences, all analy were conducted using SAT as a covariate.

Manipulation Check

To assess the initial effectiveness of the intervention probed participants for their beliefs about the malleabilit intelligence. During an unconnected study that took p not more than a week after the third session of the i vention was complete, participants in all three condition the study filled out a number of questionnaires relate academic attitudes and abilities. Embedded in these tionnaires were two items that assessed participants’ ception of intelligence (“you have a certain amount intelligence and you really can’t do much to change it”; can learn new things, but you can’t really change your b intelligence“). Both were measured on 6-point scales chored at the endpoints by the phrases strongly agre and strongly disagree (6). Participants’ responses o

, AND GOOD

items were highly correlated (r 5 .84), so anindex of malleability was formed by computing their mean. The

or pal nce ri- a

d pen

ol

tion e

ma st in mal-

nge tly, ste- and sse

s an aca- inter aca-

sta rtic-

brie t (Af g a

tion. rent dents cale ce of tion, uni- lete the

- ions had ici- ighly

he ce:

e is of The ce, ,

n nly ence ge of pals’ ndi- an as ugh the

lleabl mics, more Mea e .05 level.

index was submitted to a 2 (race: African American Caucasian)3 3 (condition: malleable pen pal, pen control, or non pen pal control) analysis of covaria (ANCOVA), which yielded a significant effect of expe mental condition,F(2, 72) 5 6.014, p , .005, and nonsignificant effect of race, (p . .29). Theinteraction did not reach significance (p . .93). Participants reporte viewing intelligence as more malleable in the malleable pal condition (M 5 4.92) than in the pen pal contr condition (M 5 4.24),t(73) 5 2.07,p , .05. Theratings of the participants in the non pen pal control condi (M 5 3.93) did notdiffer significantly from those of th intervention control participants (M 5 4.24). Thus, the intervention appears to have successfully altered the leable pen pals’ views in the predicted direction, at lea the short term, and to have left the beliefs about the leability of intelligence intact among control pen pals.

Long-Term Effects

Our chief concern, however, was whether these cha attitudes would hold over time and, more importan whether they would influence participants’ reactions to reotype threat and improve their academic attitudes performance. Two sets of measures were obtained to a the effectiveness of the intervention—attitude measure official grade transcripts collected at the end of the demic year (near the beginning of June). Because the vention had taken place in the Winter quarter of the demic year, approximately 9 weeks passed between the of the intervention and the final measurement of the pa ipants’ attitudes.

Attitudes at year’s end were obtained by means of a telephone interview conducted by a research assistan rican American female) who was purportedly conductin

TA Short-Term and Long

Measure

Malleable pen pal

Blacks (n 5 16)

Whites (n 5 12)

Short-term malleability beliefs 5.04a 4.81ac

Long-term malleability beliefs 5.42a 4.70ab

Enjoy academics 4.38b 5.43ad

Academics are important 4.77b 5.61a

Perceived stereotype threat 5.22a 1.62b

Spring quarter GPA 3.32a 3.55a

Note.Higher values indicate stronger belief that intelligence is ma stereotype threat, and higher GPA. All means are adjusted by SAT. Means sharing a common superscript do not differ.

STEREOTYPE THREAT AN

survey of attitudes about the academic life at Stanford. The interviewer made no mention of the pen pal program, and

l-

d

ss d

-

rt

f -

none of the participants inquired if there was a connec In addition to a number of filler questions (about cur university policies, current events, etc.), she asked stu to answer several questions—in the form of Likert s statements—aimed at measuring students’ experien stereotype threat, their degree of academic identifica and their enjoyment of the educational process at the versity. Grade transcripts from the first available comp grading period (spring quarter) were obtained from registrar later in the summer.

Beliefs about the malleability of intelligence.Embed ded in the set of interview items were the two quest regarding the malleability of intelligence participants initially answered shortly after the intervention. Part pants’ responses on these items were once again h correlated (r 5 .85) andthus were averaged to form t malleability index. The index was submitted to a 2 (ra African American or Caucasian)3 3 (condition: malleabl pen pal, pen pal control, or non pen pal control) analys covariance (ANCOVA) using SAT as the covariate. ANCOVA yielded significant main effects of both ra F(1, 72) 5 6.03, p , .02, andexperimental condition F(2, 72) 5 19.638,p , .0001, but norace by conditio interaction (F , 1). The results (Table 1) suggest not o that the attitude change created by the malleable intellig intervention endured, but also it appears that the passa time widened the differences between malleable pen beliefs and those of participants in the two control co tions.2 Interestingly, there was a tendency for Afric Americans in all three conditions to view intelligence more malleable than did their White counterparts, tho the difference only reached marginal significance in

2

1 Effects of Intervention

Experimental condition

Control pen pal Non pen pal control

Blacks (n 5 12)

Whites (n 5 11)

Blacks (n 5 14)

Whites (n 5 14)

4.40ab 4.07ab 4.15b 3.72b

4.31bc 3.79cd 3.91cd 3.20d

3.47c 4.89ab 3.42c 5.81d

3.89c 5.67ab 3.45c 5.71a

4.70a 1.42b 5.17a 1.26b

3.05b 3.34ac 3.10bc 3.35a

e, greater enjoyment of academics, greater identification with acade ns within rows not sharing a common superscript differ at least at th

119EORIES OF INTELLIGENCE

BLE -Term

D TH

All analyses were initially conducted including participant gender as a factor. Because no main or interaction effects were found, the gender variable will not be discussed further.

find 97). the

ffers pe a ss). e ss— rd?” on

ngs

n, re ican

les ast

his i lled

as a the

tici- oces

of trol, t

tive een was d trol

tify uce hey ses ues ake rts, nt? mic

per- ace t

ics emic rne, ndi- emic

nom- (Os- Our uld sup- ics ter-

ion, e ca-

- nts par- d to ikert l., in a ste- ents

udg- es”). raged er- fect

can

t mic tu- not

ce of s. It ing

ment,

n and

ieve- as a

heir ipts. the f

, be-

RIED

malleable pen pal condition,t(73) 5 1.91, p , .06. This finding of a race difference on this measure replicates ings from earlier research (e.g., Aronson & Damiani, 19 African Americans appear to be more likely to endorse incremental theory of intelligence, perhaps because it o a self-protective shield against the intelligence stereoty well as the tendency to underperform (Aronson, in pre

Enjoyment of the educational process.Students wer asked “How much do you enjoy the educational proce studying, going to class, taking tests, etc.—at Stanfo and were asked to indicate their degree of enjoyment 7-point scale, with 1 indicatinglow and 7 indicatinghigh enjoyment. The ANCOVA performed on these rati yielded a significant effect of race,F(1, 73) 5 37.10,p , .0001, asignificant effect of condition,F(2, 73) 5 3.43, p , .05, and asignificant race by condition interactio F(2, 73) 5 4.23, p , .02. The adjusted means a presented in Table 1. Thus in general, African Amer students in this sample reported enjoying academics than their White counterparts, a finding in line with p ethnographic research (e.g., Feagin & Sikes, 1994). T somewhat striking, considering the fact that we contro for ability and preparedness by using SAT scores covariate. But the intervention appeared to moderate race differences in enjoyment. African American par pants indicated greater enjoyment of the educational pr if they had written letters advocating the malleability intelligence than if they had been in the pen pal con t(72) 5 2.05, p , .05. African Americans’ enjoymen ratings did not differ across the two control conditions (t 5 .12). White participants’ ratings also showed a posi effect of the intervention, although the difference betw the malleable pen pal condition and the pen pal control not significant (p 5 .15). Interestingly, Whites reporte significantly higher enjoyment in the non pen pal con than did their counterparts in the pen pal control,t(72) 5 2.62, p , .02.

Identification with academic achievement.We pre- dicted that participants would be less likely to disiden with academic achievement—less likely, that is, to red the centrality of academics to their self-concepts—if t were convinced that intelligence was malleable. To as this, participants were asked to answer the following q tion: “Considering all the things that matter to you and m you who you are (e.g., friends, family, activities, spo talents, etc.), how important is academic achieveme Participants were asked to rate their valuation of acade on a 7-point scale, with 1 indicatinglow and 7 indicating high importance. The ANCOVA (as described above) formed on this measure yielded a significant effect of r F(1, 72) 5 32.76, p , .0001, amarginally significan effect of experimental condition,F(2, 72) 5 2.81, p ,

120 ARONSON, F

.07, and asignificant race3 condition interaction,F(2, 72) 5 4.10,p , .05. Thepattern of adjusted means (Table

-

s

a

s

s

s

s -

” s

,

1) is similar to that found for the enjoyment-of-academ question. Consistent with previous accounts of the acad identification of various ethnic groups (e.g., Osbou 1995), African Americans across all experimental co tions appeared to base their self-worth less upon acad achievement than did their White counterparts, a phe enon believed to stem in part from stereotype threat bourne, 1995; Steele, 1997; Steele & Aronson, 1995). prediction that a malleable theory of intelligence co attenuate this divestment of self from academics was ported. African-Americans reported valuing academ more in the malleable pen pal condition than their coun parts in the pen pal control condition,t(72) 5 2.22, p , .05, or than those in the non pen pal control condit t(72) 5 3.85, p , .001. For White participants, th intervention apparently had little effect on their identifi tion with academics (allt’s, ns).

Perceptions of stereotype threat.Did these improve ments in enjoyment and identification for Black stude result from a direct reduction in stereotype threat? Ap ently not. During the interview, participants were aske indicate their degree of agreement (again on 7-point L scales) with two items past research (e.g., Steele et a press) has used to measure students’ perceptions of reotype threatening environment (“people make judgm about my abilities based on my race,” people make j ments about my racial group based on my performanc These items were highly correlated and thus were ave to form an index of stereotype threat. The ANCOVA p formed on the index revealed only a highly significant ef of participant race,F(1, 72) 5 196.76, p , .0001. Regardless of experimental condition, African Ameri participants reported more stereotype threat (M 5 5.40) than did White participants (M 5 1.46). Thus, in contras to other interventions, which have lifted the acade achievement and identification of African American s dents (e.g., Steele et al., in press), this intervention did appear to do so by reducing students’ direct experien feeling judged by others through the lens of stereotype is therefore likely that the intervention worked by chang their responses to a stereotype threatening environ rather than their direct perception of it.

Academic performance.The most important questio was whether the change in attitudes about intelligence toward academics helped facilitate actual gains in ach ment. To assess whether participants’ grades improved result of participating in the intervention, we computed t grade point averages from their official grade transcr These were submitted to the ANCOVA, yielding only main effects of race,F(1, 72) 5 9.62, p , .01, and o experimental condition,F(2, 72) 5 4.93, p , .01. There are several striking features of this data pattern. First

, AND GOOD

cause we controlled for SAT, the race difference in aca- demic performance is remarkable. SAT was a highly sig-

, sis the

ican our 995 mer whic xam fro

ex a d a

by ng t e of

l the for ell. ows the pen

nce tha

t. Fo nly

pre-

n

A. its

t, we n the nce d to the ed a s as ition

ut irect ility see a ility ct of ak- e our bility itive ted

les as at the bles. per-

rm m iated with gle a cance levels. level e mediator.

nificant predictor of participant GPA,F(1, 72) 5 36.28 p , .0001. Nonetheless, controlling for it in the analy did not eliminate the race gap. In each condition of experiment Whites obtained higher grades than Afr Americans. Although this is certainly consistent with reasoning (e.g., Steele, 1997; Steele & Aronson, 1 about the extra academic burdens faced by African A icans, it is nonetheless surprising to see the degree to race appears to influence academic performance. To e ine whether the observed race gap may have stemmed stereotype threat, we included the stereotype threat ind a factor in the above described ANCOVA, which yielde less strong, but still significant effect of race,F(1, 71) 5 4.80, p , .05. Thus, some other factor not captured SAT or our measures of stereotype threat were operati depress African Americans’ grades relative to thos Whites.

The second striking feature of these data is how wel malleable intelligence intervention worked—not only African Americans, but for the White participants as w As inspection of the adjusted means in Table 1 sh African Americans tended to obtain higher grades in malleable pen pal condition, both compared to the no pal control participants,t(72) 5 2.19, p , .05, andthose in the pen pal control,t(72) 5 2.24, p , .05. This strongly suggests that it was the malleability-of-intellige message—not some other feature of the intervention— was responsible for the gains in academic achievemen the White participants, the condition difference was o marginally significant, but in the same direction as

FIG. 1. Standardized direct effects of malleability training, long-te a double arrow indicate correlations, and values associated with a sin Bracketed values reflect standardized direct effects and significance

STEREOTYPE THREAT AN

dicted; higher grades were obtained by participants in the malleable pen pal conditions than in either the pen pa

) - h -

m s

o

,

t r

control condition,t(72) 5 1.76,p , .09, or in the non pe pal control condition,t(72) 5 1.82, p , .08.

Long-term malleability beliefs as mediator of GP3

Because our intervention seemed to work so well, in effects on both attitudes and academic achievemen were interested in additional data that might strengthe argument that changes in implicit theories of intellige underlay these improvements. Specifically, we wante determine if malleability of intelligence beliefs mediated differences we found in GPA. To this end, we conduct mediational analysis using long-term malleability belief the mediator in contrasting the malleable pen pal cond with the two control conditions.

As Fig. 1 shows, malleability training and SAT (witho the presence of the mediator) both had significant d effects on GPA. To conclude that long-term malleab beliefs mediated this relationship, we should expect to reduction in the standardized direct effect of malleab training on GPA. However, the standardized direct effe malleability training was actually strengthened, not we ened, in the presence of the mediator. Thus, becaus analyses did not show a reduced direct effect of mallea training on GPA, we cannot conclude that the pos effects of the malleability training on GPA were media by malleability beliefs.4

3 Mediation analyses were also conducted using the attitude variab dependent variables. In all cases, the results failed to indicate th malleability beliefs mediated the intervention’s effects on these varia

4 Similar results were obtained when the mediation analysis was

alleability beliefs, and SAT on GPA. Note: In the model, values assoc rrow indicate standardized direct effects and the corresponding signifi s of malleability training and SAT on GPA without the presence of th

121EORIES OF INTELLIGENCE

D TH

l

formed using only Black subjects. All paths were significant except the path from SAT to Long-Term Malleability Beliefs,B 5 .07, p . .53, and the path from Long-Term Malleability Beliefs to GPA,B 5 2.236, p .

ins cor- this e e 2

lysis ely

AT ntel- tes

ntel- one nce in t

hich naly ang cif- .25

sible the

th tion

inst ter- tha

the larly

ces- hout ffer- cces- the

veral ced

ore the

uld cap-

rag- eri-

hree nce tion

ical take l, it have n as

on. d by that have Aron- n to mal- the

hus

long-term ter the end of n Pen Pal

Why did this analysis fail to support our explain the ga in GPA found in the malleable pen pal condition? Inter relations (Table 2) were computed to shed light on issue. First, the mediation analysis revealed a negativ fect of long-term malleability beliefs on GPA. As Table shows, the negative effects found in the mediation ana may reflect the general tendency for SAT to positiv correlate with GPA and further, for people with low S scores to more strongly endorse a malleable view of i ligence. This tendency for people with low standardized scores to more strongly endorse a malleable view of i ligence may be a protective strategy— believing that can get smarter when faced with poor prior performa preserves a sense of hope that increased performance future is possible (Aronson, in press).

Second, in addition to the rather small sample size, w can make intercorrelations unstable, the mediational a sis may also have been frustrated by a restricted r problem with the malleability of intelligence scale. Spe ically the mean on this measure for all subjects was 4 with a standard deviation of 1.17 and a maximum pos score of 6. And most importantly, for participants in malleable pen pal condition, the means approached maximum possible rating, suggesting that the manipula created a ceiling effect, which would certainly work aga finding supportive mediational results. One possible in pretation is that most participants expressed a belief intelligence is more malleable than fixed—when asked.But, because of the nature of the intervention, those in malleable pen pal conditions may have been particu

TA Table of Correlati

Black/White

Short-term malleability

beliefs

Long-term malleability

beliefs Enjoy

academ

Short-term malleability beliefs .855*** 2.007 Long-term malleability beliefs .811*** .098 Enjoy academics .206 .317* Academics are important .340* .562*** .592* Perceived stereotype threat .165 .077 .1 Spring quarter GPA 2.164 2.066 2.169 SAT 2.254 2.160 2.373* Malleability training .379* .596*** .528*

Note.The short-term malleability beliefs variable was collected no malleability beliefs, enjoyment of academics, academic importance, the intervention. The malleability training variable was coded as fol Control Condition, 0.

* p , .05. ** p , .01.

*** p , .001.

122 ARONSON, F

.19. Thus, Long-Term Malleability Beliefs did not mediate malleability training’s effect on GPA for Black subjects.

f-

t

he

- e

,

e

t

convinced of this, and the conviction was, moreover, ac sible enough to affect their achievement behaviors, wit our measures being sufficient to capture this critical di ence. Indeed, these differences in mere belief versus a sible conviction (see Fazio, 1995) may help explain seemingly inconsistent result found here and in se other studies showing Blacks to be not only more convin than Whites of the malleability of intelligence but also m prone to academic failure. Further studies investigating mediating effects of malleability training on GPA sho include measures of attitude accessibility designed to ture such differences.

Although the mediation results were less than encou ing, it is hard to ignore both the current and past exp mental support for the hypothesis. The results of t experiments have shown that malleability-of-intellige manipulations both boost performance and identifica relative to control groups that did not receive this crit element of the manipulation. We are thus inclined to the lack of mediational support with a grain of salt. Stil is intriguing to speculate about other factors that may driven these gains that could conceivably have arise artifacts of the malleability-of-intelligence manipulati One such factor could be increased motivation produce the combination of stereotype threat with the belief intelligence is malleable. Simply put, perhaps as we seen in past stereotype threat studies (e.g., Steele & son, 1995), stereotype threat boosts one’s motivatio disprove the stereotype. The belief that intelligence is leable may act to sustain this motivation by creating conviction that one’s efforts will amount to real gains. T

2 mong the Variables

Academics are important

Perceived stereotype

threat Spring

quarter GPA SAT Malleability

training

.120 .033 2.153 2.285 .480**

.138 .136 2.113 2.405* .618*** 2.156 2.067 2.298 2.412* .130

.066 2.233 2.405* .085 .032 .182 .087 .240

2.108 .037 .577*** .079 2.353* 2.148 .521*** 2.395*

.606*** .117 .080 2.368*

e than a week after the end of the intervention. Variables measuring erceived stereotype threat were collected approximately 9 weeks af Malleable Pen Pal Condition, 1; Pen Pal Control Condition, 0; No

, AND GOOD

BLE ons a

ics

** 72

**

t mor and p lows:

RIED

it is possible that, by itself, the malleability-of-intelligence belief may have a less powerful influence on GPA than a

but hy-

ex- just nce, own

their thei hey they f the the

, ov more hite is i ear can less the

ved, en

stor ence ility d si nti- and

lass that his e di am- pas

y is lso ts th feri- eck

er- y of ade s on

from rade ni-

nce s out the

eport per-

the gs reo- ally ave ched d, it thing

tereo- f the hreat could

eve-

ed the

e less 95), s by omy

ican erty

on omic ent

omic in

The ge-

id- igher elop job di-

than n,

ven e per- y stu-

y to stra- and

lum,

D TH

combination of the belief coupled with a threatening— motivating—stereotype. We are currently testing this pothesis with additional experiments.

GENERAL DISCUSSION

The results of this intervention were on the whole tremely encouraging. African American students, after three sessions of advocating the malleability of intellige created an enduring and beneficial change in their attitudes about intelligence. This change improved academic profile to a significant degree: compared to counterparts in either of the two control conditions, t reported enjoying and valuing academics more and received higher grades. The intervention had some o same positive effects for White students, though not to same degree. One clear difference was that whereas time, African American students appeared to become convinced of the expandability of intelligence, the W students’ attitude change did not persist. Perhaps th related to the greater baseline endorsement Blacks app give to the incremental view, which we have suggested stem as much from desires as convictions. Nonethe White students seemed to enjoy some benefit from malleable intelligence intervention; their grades impro though their reported academic identification and enjoym did not.

At the same time we must note the less encouraging these results tell us about the African American experi in college. Even after controlling for preparation and ab (as measured by SAT scores), these students receive nificantly lower grades, showed significantly lower ide fication or engagement with the schooling process, reported enjoying themselves less than their White c mates. This finding—as well as the additional finding controlling for stereotype threat did not fully eliminate t gap in performance and engagement— underscores th ficulty these students face on predominantly White c puses. Our findings, moreover, are consistent with findings in suggesting that at least part of this difficult created by suspicions of intellectual inferiority. And, a consistent with past research, the present study sugges such doubts may be particularly damaging when the in ority can be seen as irremediable (Aronson, 1999; Dw 1999).

It is interesting to note that this relatively simple int vention of changing students’ views of the expandabilit intelligence had about as much positive influence on gr as some larger scale, multifaceted interventions, such a reported by Steele et al. (in press), who, also working the stereotype threat model, successfully raised the g (by four tenths of a grade) of Black freshmen at the U

STEREOTYPE THREAT AN

versity of Michigan. This program used a combination of special recruitment procedures, weekly study groups, an

r

er

s to

,

t

y

g-

-

f-

t

at

,

s e

s

frequent mastery workshops. There is a critical differe between these two approaches that, we believe, hold hope that even more dramatic gains could occur if approaches were combined. Specifically, Steele et al. r that their gains were mediated by a direct effect on ceived stereotype threat. That is, their intervention had effect of significantly changing Black students’ feelin about being perceived by the larger community in a ste typical way—they felt less looked down upon academic by their White peers. Our intervention clearly did not h this effect; perceptions of stereotype threat were untou by the conceptions-of-intelligence manipulation. Instea appeared that the attitude change intevention did some to change their academically relevant responses to s type threat. It is quite possible then, that a marriage o two approaches—which could both reduce stereotype t and reduce some of the negative responses it spurs— have an additive effect, boosting African American achi ment more than either approach alone.

Implications

The quality of life for Black Americans has improv dramatically in the past few decades. Discrimination on basis of race has diminished— or, at least, has becom blatant (e.g., Kinder, 1986; Pettigrew & Meertens, 19 and equal access is guaranteed by law, if not alway practice. These changes, along with a surging econ have helped make the current times better for Afr Americans than ever, with fewer unemployed or in pov than ever and the highest level of health and optimism record (Cose, 1999). Still, despite this brightening econ and cultural picture, African American school achievem stubbornly lags that of Whites. In one sense the econ forces that have improved the lives of all minorities America make the educational disparities more dire. economy has become and is likely to remain “knowled driven”; making a living increasingly requires most indiv uals, regardless of race or gender, not only to pursue h education, but to draw fully upon its resources to dev the kind of skills needed to compete and thrive in the market (Hershberg, 1998; Murnane & Levy, 1997). In viduals unable to attend or finish college are, more ever, at risk of being left behind (Fullilove & Treisma 1990). The apparent irreversibility of the knowledge-dri economy underscores the importance of addressing th sistent underachievement of underrepresented minorit dents at all levels of schooling.

We think that shaping students’ conceptions of abilit promote more adaptive responses to the inevitable fru tions and threats posed by racially integrated colleges schools could easily become part of school curricu

123EORIES OF INTELLIGENCE

d perhaps as a complement to other structure-altering ap- proaches, such as cooperative learning (e.g., Aronson &

ation ul in sum olers their

an oris ool

agin no- (e.g ude ave suc tern

999) s, & ong- and tion lson e o lay

with

tion.

ble rk

disi- -

ut

hil- t

ele, cient gy,

he

d aylor

01). type

n and -

ired- -

.

and gy,

ity,

h to

ture beliefs.

termi- . Petty cees.

der- nves- d

k

sso-

ong ley:

es.

di,

he

tion.

k,

d

RIED

Patnoe, 1997) and other forms of intergroup communic (Steele et al., in press) that have also proved usef addressing minority student underachievement. Pre ably, shaping or reshaping the views of grade-scho would be easier than with college students, since attitudes about intelligence may be less entrenched because young children tend start out as malleable the and grow more entitivistic as they move through sch (Dweck, 1999).

In the past 2 decades, there has been encour progress within the scientific community. Traditional tions of what creates academic and life success Hernnstein & Murray, 1994) are broadening to incl factors other than innate intellectual ability. Scholars h begun to recognize and demonstrate the importance of factors as emotional regulation (e.g., Goleman, 1995; S berg, 1996b), self-theories and goals (e.g., Dweck, 1 and explanatory styles (e.g., Nolen-Hoeksema, Girgu Seligman, 1987). They have also begun rethinking l standing and widespread notions of what intelligence is in particular, how it can expand in response to interven (e.g., Gardner, 1983; Sternberg, 1998; Thompson & Ne 2001). The current study suggests the potential valu making these modern views as well known to the population as they are to those in academia.

REFERENCES

Aronson, E., & Patnoe, S. (1997).The jigsaw classroom.New York: Longman.

Aronson, J. (in press). Stereotype threat: Contending and coping unnerving expectations. In J. Aronson & D. Cordova (Eds.),Improving education: Classic and contemporary lessons from psychology.New York: Academic Press.

Aronson, J. (1997).The effects of conceptions of ability on task valua Unpublished manuscript, New York University.

Aronson, J. (1999).The effects of conceiving ability as fixed or improva on responses to stereotype threat.Unpublished manuscript, New Yo University.

Aronson, J., Blanton, H., & Cooper, J. (1995). From dissonance to dentification: Selectivity in the self-affirmation process.Journal of Per sonality and Social Psychology,58, 1062–1072.

Aronson, J., & Damiani, M. (1997).Ethnic differences in beliefs abo intelligence.Unpublished manuscript University of Texas, Austin.

Aronson, J., & Good, C. (2000).Development and measurement of c dren’s vulnerability to racial and gender ability stereotypes.Manuscrip in preparation.

Aronson, J., Lustina, M., Good, C., Keough, K., Brown, J. L., & Ste C. M. (1999). When White men can’t do math: Necessary and suffi factors in stereotype threat.Journal of Experimental Social Psycholo 35, 11–23.

Aronson, J., Quinn, D., & Spencer, S. (1998).Stereotype threat and t academic underperformance of women and minorities.In J. Swim & C. Stangor (Eds.),Stigma: The target’s perspective.New York: Academic Press.

124 ARONSON, F

Bigler, R. S. (1999). Psychological interventions designed to counter sexism in children: Empirical limitations and theoretical foundations. In

-

d ts

g

.,

h - ,

,

, f

W. B. Swann, J. H. Langlois, & L. A. Gilbert (Eds.),Sexism an stereotypes in modern society: The gender science of Janet T Spence.Washington, DC: American Psychological Association.

Blascovich, J., Spencer, S. J., Quinn, D. M., & Steele, C. M. (20 African Americans and high blood pressure: The role of stereo threat.Psychological Science,12, 225–229.

Cacioppo, J. T., & Petty, R. E. (1979). Effects of message repetitio position on cognitive response, recall, and persuasion.Journal of Per sonality and Social Psychology,37, 97–109.

Cook, T. D., & Wadsworth, A. (1972). Attitude change and the pa associates learning of minimal cognitive elements.Journal of Person ality, 40, 50 – 61.

Cose, E. (1999, June 7). The good news about Black America.Newsweek

Devine, P. G. (1989). Stereotypes and prejudice: Their controlled automatic components.Journal of Personality and Social Psycholo 56, 5–18.

Dweck, C. S. (1986). Motivational processes affecting learning.American Psychologist,41, 1040 –1048.

Dweck, C. S. (1999).Self-theories: Their role in motivation, personal and development.Philadelphia, PA: Taylor & Francis.

Dweck, C. S., & Leggett, E. L. (1988). A social cognitive approac motivation and personality.Psychological Review,95, 256 –273.

Eccles, J. S., & Wigfield, A. (1995). In the mind of the actor: The struc of adolescents’ achievement task values and expectancy related Personality and Social Psychology Bulletin,21(3), 215–225.

Fazio, R. H. (1995). Attitudes as object-evaluation associations: De nants, consequences, and correlates of attitude accessibility. In R & J. Krosnick (Eds.),Attitude strength: Antecedents and consequen Hillsdale, NJ: Erlbaum.

Fazio, R. H., & Williams, C. J. (1986). Attitude accessibility as a mo ator of the attitude-perception and attitude-behavior relations: An i tigation of the 1984 presidential election.Journal of Personality an Social Psychology,51, 504 –514.

Feagin, J. R., & Sikes, M. P. (1994).Living with racism: The Blac middle-class experience.Boston: Beacon.

Fried, C. B., & Aronson, E. (1995). Hypocrisy, misattribution, and di nance reduction.Personality and Social Psychology Bulletin,21, 925– 933.

Fullilove, R., & Treisman, P. (1990). Mathematics achievement am African American undergraduates the University of California Berke An evaluation of the mathematics workshop program.Journal of Negro Education,59, 463– 479.

Gardner, H. (1983).Frames of mind: The theory of multiple intelligenc New York: Basic Books.

Garibaldi, A. M. (1991). Blacks in college. In C. Willie, A. M. Garibal & W. L. Reed (Eds.),The education of African Americans.New York: Auburn House.

Goleman, D. (1995).Emotional intelligence.New York: Bantam.

Good, C., Aronson, J., & Harder, J. A. (2000).Stereotype threat and t kernal of truth.Unpublished manuscript, University of Texas.

Gould, S. J. (1981).The mismeasure of man.New York: Norton.

Harter, S. (1990).Processes underlying adolescent self-concept forma In R. Montmayor, G. R. Adams, & T. P. Gullotta (Eds.),From childhood to adolescence: A transitional period?(pp. 205–239). Newbury Par CA: Sage.

Herrnstein, R. J., & Murray, C. (1994).The bell curve: Intelligence an class structure in American life.New York: Free Press.

Hershberg, T. (1998). The case for new standards in education.Education

, AND GOOD

Week,17, 1–2.

Hewstone, M. (1996). Contact and categorization: Social psychological

, C.

of ribed

lli- ation

gat- udes

: An st

ance

cho-

998). : The y

ci-

(Ed.)

, N., J., & -

sub- ,

rned

at the

rgy al

self-

am- :

ice in

for ds.), n,

self- iefing .

-

al

ans.

ctual

ctual d

oem, t: A

ntelli-

d the

D TH

interventions to change intergroup relations. In C. N. Macrae Stangor, & M. Hewstone (Eds.),Stereotypes and stereotyping.New York: Guilford.

Higgins, E. T., & Rholes, W. S. (1978). “Saying is believing”: Effects message modification on memory and liking for the person desc Journal of Experimental Social Psychology,14, 363–378.

Hong, Y., Chiu, C., & Dweck, C. S. (1995). Implicit theories of inte gence: Reconsidering the role of confidence in achievement motiv In M. H. Kernis et al. (Eds.),Efficacy, agency, and self-esteem(pp. 197–216). New York: Plenum.

Huesman, L., Eron, L., Klein, R., Brice, P., & Fischer, P. (1983). Miti ing the imitation of aggressive behaviors by changing chilren’s attit about violence.Journal of Personality and Social Psychology,44, 899 – 910.

Jacoby, R., & Glauberman, N. (1995).The bell curve debate.New York: Random House.

Jencks, C., & Phillips, M. (1998). The Black–White test score gap introduction. In C. Jencks & M. Phillips (Eds.),The Black–White te score gap.Washington DC: Brookings Instit.

Kinder, D. R. (1986). The continuing American dilemma: White resist to racial change forty years after Myrdal.Journal of Social Issues,42, 151–171.

Lewontin, R. C., Rose, S., & Kamin, L. J. (1984).Not in our genes.New York: Pantheon.

Major, B., & Schmader, T. (1998). Coping with stigma through psy logical disengagement. In J. Swim & C. Stangor (Eds.),Stigma: The target’s perspective.New York: Academic Press.

Major, B., Spencer, S., Schmader, T., Wolfe, C., & Crocker, J. (1 Coping with negative stereotypes about intellectual performance role of psychological disengagement.Personality & Social Psycholog Bulletin, 24(1), 34 –50.

Murnane, R. J., & Levy, F. (1997).Teaching the new basic skills: Prin ples for educating children to thrive in a changing economy.New York: Free Press.

Neisser, U. (1986). New answers to an old question. In U. Neisser The school achievement of minority children.Hillsdale, NJ: Erlbaum.

Neisser, U., Boodoo, G., Bouchard, T. J., Jr., Boykin, A. W., Brody Ceci, S. J., Halpern, D. F., Loehlin, J. C., Perloff, R., Sternberg, R. Urbina, S. (1996). Intelligence: Knowns and unknowns.American Psy chologist,51(2), 77–101.

Nicholls, J. (1984). Achievement motivation, conceptions of ability, jective experience, task choice, and performance.Psychological Review 91, 328 –346.

Nolen-Hoeksema, S., Girgus, J. S., & Seligman, M. E. (1987). Lea

STEREOTYPE THREAT AN

helplessness in children: A longitudinal study of depression, achieve- ment, and explanatory style.Journal of Personality & Social Psychol- ogy, 51, 435– 442.

.

.

,

Osbourne, J. W. (1995). Academics, self-esteem, and race: A look underlying assumptions of the disidentification hypothesis.Personality and Social Psychology Bulletin,21, 449 – 455.

Pallak, Cook, D. A., & Sullivan, J. (1981). Commitment and ene conservation. In L. Bickman (Ed.),Applied social psychology annu (Vol. 1, pp. 15– 40). Beverly Hills, CA: Sage.

Pelham, B. W., & Swann, W. B. (1989). From self-conceptions to worth: On the sources and structure of global self-esteem.Journal of Personality & Social Psychology,57, 672– 680.

Pettigrew, T. F. (1981). Extending the stereotype concept. In D. L. H ilton (Ed.), Cognitive processes in intergroup behavior.Hillsdale, NJ Erlbaum.

Pettigrew, T. F., & Meertens, R. W. (1995). Subtle and blatant prejud Western Europe.European Journal of Social Psychology,25, 57–75.

Petty, R. E., & Wegener, D. T (1998). Attitude change: Multiple roles persuasion variables. In D. T. Gilbert, S. T. Fiske, & G. Lindzey (E Handbook of social psychology(3rd ed., Vol. 1, pp. 323–329). Bosto MA: McGraw–Hill.

Ross, L., Lepper, M. R., & Hubbard, M. (1975). Perseverance in perception and social percepion: Biased attribution in the debr paradigm.Journal of Personality and Social Psychology,35, 880 – 892

Schwartz, B. (1997). Psychology, idea technology, and ideology.Psycho logical Science,8(1), 21–27.

Smith, T. W. (1990).Ethnic images.GSS Topical Report No. 19. Nation Opinion Research Center.

Steele, C. M. (1992, April). Race and the schooling of Black Americ The Atlantic Monthly.

Steele, C. M. (1997). A threat in the air: How stereotypes shape intelle identity and performance.American Psychologist,52, 613– 629.

Steele, C. M., & Aronson, J. (1995). Stereotype threat and the intelle test performance of African-Americans.Journal of Personality an Social Psychology,68, 797– 811.

Steele, C. M., Spencer, S. J., Hummel, M., Carter, K., Harber, K., Sch D., & Nisbett, R. (in press). African-American college achievemen “wise” intervention.Harvard Educational Review.

Sternberg, R. J. (1996a). Myths, countermyths, and truths about i gence.Educational Researcher,25, 11–16.

Sternberg, R. J. (1996b).Successful intelligence.New York: Simon & Schuster.

Sternberg, R. J. (1998). Abilities are forms of developing expertise.Edu- cational Researcher,27, 11–20.

Thompson, R. A., & Nelson, C. A. (2001). Developmental science an media: Early brain development.American Psychologist,56, 5–15.

125EORIES OF INTELLIGENCE

Utman, C. H. (1997). Performance effects of motivational state: A meta ananlysis.Personality and Social Psychology Review,1, 170 –182.