With a Little Help From My Cross-Group Friend: Reducing Anxiety in

Intergroup Contexts Through Cross-Group Friendship

Elizabeth Page-Gould and Rodolfo Mendoza-Denton

University of California, Berkeley

Linda R. Tropp

University of Massachusetts at Amherst

The authors induced cross-group friendship between Latinos/as and Whites to test the effects of

cross-group friendship on anxiety in intergroup contexts. Cross-group friendship led to decreases in

cortisol reactivity (a hormonal correlate of stress; W. R. Lovallo & T. L. Thomas, 2000) over 3 friendship

meetings among participants high in race-based rejection sensitivity (R. Mendoza-Denton, G. Downey,

V. J. Purdie, A. Davis, & J. Pietrzak, 2002) and participants high in implicit prejudice (A. G. Greenwald,

B. A. Nosek, & M. R. Banaji, 2003). Cross-group partners’ prior intergroup contact moderated the

relationship between race-based rejection sensitivity and cortisol reactivity. Following the manipulation,

participants kept daily diaries of their experiences in an ethnically diverse setting. Implicitly prejudiced

participants initiated more intergroup interactions during the diary period after making a cross-group

friend. Participants who had made a cross-group friend reported lower anxious mood during the diary

period, which compensated for greater anxious mood among participants high in race-based rejection

sensitivity. These findings provide experimental evidence that cross-group friendship is beneficial for

people who are likely to experience anxiety in intergroup contexts.

Keywords: cross-group friendship, intergroup anxiety, close relationships, intergroup contact, intergroup

relations

Even though interactions between members of different social

groups are sometimes characterized by anxiety and threat (Blas-

covich, Mendes, Hunter, Lickel, & Kowai-Bell, 2001; Mendes,

Blascovich, Lickel, & Hunter, 2002; Stephan & Stephan, 1985,

2000), a growing body of research suggests that cross-group

friendship can attenuate such anxiety. Cross-group friendship is

associated with low levels of self-reported intergroup anxiety

(Paolini, Hewstone, Voci, Harwood, & Cairns, 2006), and the

improvements in intergroup attitudes that result from intergroup

contact seem especially evident when such contact is characterized

by friendship (Pettigrew & Tropp, 2005, 2006, footnote 4; Tropp,

2007). Longitudinal research on the acquisition of cross-group

friends in college (Levin, van Laar, & Sidanius, 2003) revealed

that students who acquired more cross-group friends by their 2nd

year reported significantly less anxiety about interacting with

outgroup members at the end of 4 years of college. Even among

groups with a recent history of violent conflict, such as Catholics

and Protestants in Northern Ireland, having close friends from the

other religious group predicted less anxiety about interactions with

members of that group (Paolini, Hewstone, Cairns, & Voci, 2004).

Despite the promise of these findings, the direction of causality

between cross-group friendship and anxiety in intergroup contexts

needs to be established experimentally. Wright and his colleagues

(see Wright, Aron, & Tropp, 2002; Wright, Brody, & Aron, 2005;

Wright, Ropp, & Tropp, 1998; Wright & van der Zande, 1999)

described research that provided initial evidence for the causal

effects of cross-group friendship on self-reported anxiety.

Majority-group participants were paired with either a same- or

cross-group partner for a series of friendly interactions, after which

their mood and attitudes were assessed. The results yielded reduc-

tions in self-reported anxiety and increases in closeness after each

interaction. The experimental nature of this research allowed for

causal inferences regarding the benefits of cross-group friendship,

a lack of which had characterized the literature on close intergroup

relationships (Pettigrew, 1998).

Building on the experimental paradigm used by Wright and

colleagues (Wright et al., 1998, 2002, 2005; Wright & van der

Elizabeth Page-Gould and Rodolfo Mendoza-Denton, Department of

Psychology, University of California, Berkeley; Linda R. Tropp, Depart-

ment of Psychology, University of Massachusetts at Amherst.

Elizabeth Page-Gould is now at Harvard University.

This research was supported by a grant from the University of Califor-

nia, Berkeley’s Center for the Development of Peace and Well-Being, and

this material is based upon work supported under a National Science

Foundation Graduate Research Fellowship to Elizabeth Page-Gould. Por-

tions of this research were presented at the 2006 meeting of the Society for

the Psychological Study of Social Issues and the 2007 meeting of the

Society of Personality and Social Psychology.

We are grateful to Wendy Berry Mendes for her many contributions to

the methods used in this study and O

zlem Ayduk for her conceptual and

quantitative guidance. We deeply thank the experimenters, Sarah Hirsch,

Tyler Jenkins, Jessica Lopez Jimenez, Ria Jose, Allison Lee, Thanh

“Trevor” Nguyen, Wendy Nguyen, Deepa Patel, Tecsia Ross, Melanie

Weininger, and Rosa Wong, for their exceptional work. We thank Ian

Dennis Miller for the customized infrastructure and software that imple-

mented the project. We are also grateful for the insightful comments given

by E. J. Horberg, Wayne Chan, Jennifer Goetz, M. Goldman-Flythe, Nina

Hansen, Christopher Lucas, Lindsay Shaw Taylor, and Stephen C. Wright

on previous versions of this article.

Correspondence concerning this article should be addressed to Elizabeth

Page-Gould, William James Hall 1410, 33 Kirkland Street, Cambridge,

MA 02138, or Rodolfo Mendoza-Denton, Institute of Personality and

Social Research, University of California, Berkeley, CA 94720-5050.

E-mail: [email protected] or [email protected]

2008, Vol. 95, No. 5, 1080 –1094 0022-3514/08/$12.00 DOI: 10.1037/0022-3514.95.5.1080

1080

Zande, 1999), we report a study in which friendship was induced

between same- and cross-group dyads of Latinos/as and Whites.

We extend prior research in several respects. First, we experimen-

tally examined the effects of cross-group friendship from the

perspectives of two groups (c.f. Pettigrew, 1998; Tropp & Petti-

grew, 2005b), adopting an approach that centers on individual

differences both in prejudicial attitudes and in concerns about

being the target of others’ prejudicial attitudes. Second, we incor-

porated a hormonal correlate of stress, cortisol reactivity, when

assessing participants’ anxious responses to the experimental ma-

nipulation of cross-group friendship. Physiological responses have

the benefit of not being subject to controlled processes or self-

presentation effects (Blascovich et al., 2001). Third, we examined

dyadic effects on cortisol reactivity by testing whether a friend’s

prior intergroup contact moderated the effects of cross-grou p

friendship. Finally, over a diary period following the final friend-

ship meeting, we examined anxious mood and initiation of cross-

group social interactions in an ethnically diverse setting. The

experimental procedure described in this paper represents a rela-

tively strong situation (Mischel, 1977; Snyder & Ickes, 1985) that

maximized conditions for friendship formation, as our interest was

in fostering closeness irrespective of participants’ preexisting pro-

pensities to form cross-group friendships.

Reducing Anxiety in Intergroup Contexts Through Cross-

Group Friendship

Generally, we propose that cross-group friendship improves

intergroup interactions through systematic disconfirmations of

negative expectations about intergroup experiences (Mendoza-

Denton, Page-Gould, & Pietrzak, 2006). If people who are predis-

posed to anxiety in intergroup contexts experience multiple epi-

sodes of intergroup contact with no attendant anxiety—as would

likely accompany the development of a new cross-group friend-

ship—then chronic anxious orientations to outgroup members may

be attenuated (Paolini et al., 2006). In this way, the formation of a

cross-group friendship may improve intergrou p interactions

among individuals who typically have negative intergroup expec-

tations (Mendoza-Denton et al., 2006).

Individual Differences in Anxiety in Intergroup Contexts

Our approach to addressing the causal benefits of cross-group

friendship is couched in the recognition that not everyone who

engages in intergroup interaction experiences threat and anxiety to

the same degree (Britt, Boniecki, Vescio, Biernat, & Brown, 1996;

Mendoza-Denton, Downey, Purdie, Davis, & Pietrzak, 2002; Plant

& Devine, 2003; Stephan & Stephan, 1984, 1985). Accordingly,

we focus on two processes that prior research and theory have

identified as relevant to experiencing threat and anxiety when

developing relationships across group boundaries: concerns about

being rejected on the basis of group membership (Mendoza-

Denton et al., 2002 ) and prejudiced attitudes ( Greenwald,

McGhee, & Schwartz, 1998). Although we originally conceptual-

ized these processes as being most relevant to minority and ma-

jority participants, respectively, we nevertheless examined their

predictive utility for both groups.

Concerns of Rejection in Intergroup Contexts

Given the centrality of concerns of rejection on the basis of

group membership for anxiety in intergroup contexts (Mendoza-

Denton et al., 2006; Stephan & Stephan, 2000), we used a scale of

race-based rejection sensitivity (RS-race; Mendoza-Denton et al.,

2002) to tap into Latino/a and White participants’ anticipatory

rejection concerns. Researchers have previously used this ques-

tionnaire to understand minority students’ sense of institutional

belonging and academic outcomes in predominantly White insti-

tutions (Mendoza-Denton et al., 2002; Mendoza-Denton, Pietrzak,

& Downey, in press; Pin˜a & Mendoza-Denton, 2003; Walton &

Cohen, 2007). Nevertheless, research converges on the notion that

concerns of rejection are relevant for members of both minority

(Mendoza-Denton et al., 2002; Stephan & Stephan, 1989; Shelton,

Richeson, & Salvatore, 2005) and majority groups (Goff, Steele, &

Davies, 2008; Plant & Devine, 2003; Stephan & Stephan, 2000;

Vorauer, Hunter, Main, & Roy, 2000). Therefore, we measured

race-based rejection concerns among members of both groups.

Prejudiced Attitudes

Research similarly suggests that prejudiced attitudes toward a

cross-group partner’s ethnic group should invoke anxiety in inter-

group interactions (Stephan & Stephan, 1985, 2000) and more

negative expectations for these interactions (Vorauer & Kumhyr,

2001). Consistent with this view, implicit prejudice has been found

to predict subtle manifestations of anxiety and hostility in cross-

group interactions (Dovidio, Kawakami, & Gaertner, 2002; Gaert-

ner & Dovidio, 2005), whereas egalitarianism has been linked with

more adaptive stress responses during interracial int eractions

(Mendes, Gray, Mendoza-Denton, Major, & Epel, 2007). In the

present research, we focused on implicit prejudice— here mea-

sured by the Implicit Association Test (IAT; Greenwald, Nosek, &

Banaji, 2003)—rather than explicit prejudice, given that the former

measure seems more resistant to demand characteristics (Fazio &

Olson, 2003; Nosek, 2005).

Hypothesis 1

Bringing together the above literatures, we hypothesized that

only participants who are likely to experience anxiety in intergroup

contexts (either because of RS-race or implicit prejudice) should

show signs of hormonal stress responses when they first meet a

cross-group partner, but that cross-group friendship should atten-

uate such stress responses over the course of friendship develop-

ment. As a corollary, participants who scored lower on measures

of RS-race or implicit prejudice were not expected to show such

attenuation in the cross-group condition because they should have

been less likely to exhibit hormonal stress responses in the first

place. In addition, we tested whether participants’ ethnicity (i.e.,

Latino/a or White) moderated these predicted patterns.

Moderating Effects of Partners’ Prior Intergroup Contact

An implied assumption of Hypothesis 1 is that the mere group

membership of a cross-group friend should be sufficient to atten-

uate hormonal stress responses among individuals who are predis-

posed to experience anxiety in intergroup contexts. However,

cross-group friends may themselves vary substantially in their

1081

CROSS-GROUP FRIENDSHIP

ability to navigate cross-group interactions. It seems reasonable to

anticipate that those with prior intergroup contact may be more

able to promote feelings of comfort in the interaction than out-

group members without prior intergroup experience. In line with

this view, White students with prior interracial contact have been

shown to perceive upcoming cross-race interactions as less de-

manding and exhibited more adaptive physiological responses to

stress during those interactions (Blascovich et al., 2001; Mendes et

al., 2002). Moreover, research examining barriers to the formation

of cross-group friendship demonstrated that people with greater

intergroup contact tend to be more accurate about how much

interest in friendship they convey to cross-group partners (Vorauer

& Sakamoto, 2006). Taken together, these perspectives suggest

that having a partner with greater intergroup contact should facil-

itate positive responses to cross-group friendship formation.

On the other hand, research suggests that being highly invested

in an intergroup interaction, as might be the case with experienced

partners (e.g., Emerson, Kimbro, & Yancey, 2002), can paradox-

ically lead people to appear less positive and less responsive in

intergroup encounters (Vorauer & Turpie, 2004). For individuals

who anticipate rejection in intergroup encounters, an outgroup

partner who conveys less positivity than is normative may be

parti cularly threate ning. Cons istent wit h this notion, Shelton,

Richeson, Salvatore, and Trawalter (2005) found that Black par-

ticipants actually preferred interacting for a short while with im-

plicitly prejudiced White partners more than with White partners

who were low in implicit prejudice. Together, these findings

present an alternate possibility that cross-group partners with sub-

stantial prior contact could engender threat, particularly among

individuals who are vigilant for social cues during the interaction.

Hypothesis 2

We expected that partners’ prior intergroup contact would mod-

erate the anxiety-reducing effects of cross-group friendship among

those participants most likely to experience anxiety in intergroup

contexts. On the one hand, we hypothesized that cortisol reactivity

should be the least pronounced among participants who were

predisposed to anxiety in intergroup contexts but also paired with

a cross-group partner with prior intergroup contact. On the other

hand, a series of recent findings have led to an alternate hypothesis

that participants with prior intergroup contact may engender

greater threat among outgroup partners who are vigilant for cues of

rejection in intergroup encounters. We also examined changes in

these effects over the friendship meetings.

Effects Beyond the Friendship

We were also interested in whether the formation of cross-group

friendship would have any intergroup effects beyond the specific

friendship dyad. Research and theory suggest that positive expe-

riences with a cross-group friend can have a positive impact on

intergroup relations that reaches outside the context of the friendship

(Allport, 1954; Paolini, Hewstone, Rubin, & Pay, 2004; Pettigrew &

Tropp, 2005), and we examined this possibility using two indicators.

The first is participants’ anxious mood in a diverse social environment

that is characterized by a high likelihood of intergroup interaction.

The second involves behaviors toward outgroup members, assessed

by initiation of intergroup interactions in this environment.

Anxious Mood in Diverse Contexts

Environments in which one might be the target of exclusion,

prejudgment, or differential treatment can engender an overall

sense of discomfort and anxiety (Inzlicht & Ben-Zeev, 2000;

Mendoza-Denton et al., 2002; Walton & Cohen, 2007). Presum-

ably, this discomfort occurs because potential recipients of such

rejection have less control over whether rejection actually occurs

than do the perpetrators, and thus, the threat of rejection may apply

chronically. Given that the setting of the current study was a

diverse social environment in which intergroup interactions are

likely to form a necessary part of everyday life (see the Setting

section below), we reasoned that RS-race in particular would be

associated with less overall comfort in this environment. To the

degree that positive attitudes fostered through cross-group friend-

ship may ease anxious expectations of future rejection, we tested

whether our experimental manipulation would lead to decreased

anxious mood within this particular setting.

Initiation of Cross-Group Interactions

Past research suggests that avoidance of outgroup members is a

behavioral correlate of anxiety in intergroup contexts (Islam &

Hewstone, 1993). By contrast, cross-group friendship has been

associated with greater openness to future interracial ties (Emerson

et al., 2002; Tropp & Pettigrew, 2005a). As such, we tested

whether our experimental manipulations of friendship would in-

crease subsequent approach-oriented behaviors toward outgroup

members, indexed here as initiation of cross-group interactions in

daily life. We also examined initiation of same-group interactions

to ensure the findings were specific to outgroup members.

Hypothesis 3

We expected that the effects of forming a new cross-group

friendship would persist in daily life. More specifically, we ex-

pected that the development of a cross-group friendship would

lead to more initiation of intergroup interactions during the diary

period, particularly among those who were originally predisposed

to anxiety in such interactions. We further hypothesized that par-

ticipants higher in RS-race would report more anxious mood over

the diary period but that this anxiety would be attenuated through

the development of cross-group friendship.

Overview of Present Research

To test the above hypotheses, we experimentally manipulated

friendship by randomly assigning Latino/a and White undergrad-

uates to same-group or cross-group friendship conditions. Partic-

ipants completed measures of RS-race, implicit prejudice, and

prior intergroup contact prior to the study. As in prior work by

Wright and colleagues (Wright et al., 1998, 2002; Wright & van

der Zande, 1999), same- and cross-group. Friendship dyads met

each week at the same time for 3 consecutive weeks and engaged

in a series of closeness-building tasks, modeled after the Fast

Friends procedure (Aron, Melinat, Aron, Vallone, & Bator, 1997),

which was designed to increase closeness between strangers. In

addition, salivary cortisol was assessed at the beginning and end of

each friendship meeting. Together, these data allowed us to ad-

dress Hypotheses 1 and 2. Following the final friendship meeting,

1082

PAGE-GOULD, MENDOZA-DENTON, AND TROPP

participants completed daily diaries of their social interactions and

mood. These additional data allowed us to address Hypothesis 3.

Methods

Setting

This study was conducted at the University of California, Berke-

ley (UC Berkeley) from the Fall of 2003 through the Spring of

2005. Over the course of data collection, the ethnic distribution of

the undergraduate population was as follows: 46.8% Asian, 34.4%

White, 12.0% Latino/a, 4.3% Black, and 2.5% other.

Participants and Procedure

Design

The experiment was a 2 (Friendship Condition: same or cross

group, between participants) ! 2 (Participant Ethnicity: Latino/a or

White, between participants) ! 3 (Meeting Number: first, second, and

third, within participants) mixed-factorial dyadic design.

1,2

We used

the same-group friendship condition as a control condition because it

allowed us to compare the effects of making a cross-group friend with

the effects of making a new friend without a cross-group component.

We conducted separate analyses using RS-race and implicit prejudice

as predictors. Prior intergroup contact (Islam & Hewstone, 1993) and

interpersonal rejection sensitivity (Downey & Feldman, 1996) were

included as covariates in all analyses (see the Materials section).

Recruitment and Initial Assessment

Undergraduate research assistants recruited Latino/a and White

participants from common areas on campus to participate in the

“Friendship and Adjustment to College Study” for $68, a $5 gift

certificate redeemable at a local smoothie store, and a chance to win

prizes in a lottery held every semester. Participants were informed that

the researchers were studying the effects of friendship formation on

adjustment to college, but they were not privy to the ethnic compo-

sition of the experimental conditions. Interested participants (N "

159) attended an initial information session in groups of 1 to 8 during

which they provided informed consent, shared their weekly schedules,

and completed the background measures.

Partner assignment. After the information session, we

matched participants into dyads, following the matching procedure

described by Aron et al. (1997). We randomly assigned partici-

pants to the same- or cross-group conditions and then matched

them with an eligible same-sex partner on the basis of coordinating

schedules. Although participants were not matched for similarity

on any characteristics other than sex, they were told their partner

was chosen on the basis of having a similar personality.

Attrition. Eleven participants did not complete the study (6.92%

attrition) because they either knew their assigned partner and did not

want to be re-paired, or they withdrew from the study because of time

constraints. Participants who did not complete the study were de-

briefed, thanked, and received partial compensation for the aspects of

the study they had completed. One dyad elected not to provide saliva

samples, so their data were only included in the diary analyses. In

addition, one cross-group pair was eliminated from the final sample

after they had completed the study because we later learned that one

of the participants was Asian American.

Final sample. The final sample (N " 144) consisted of 64

Latino/a participants (78% female) and 80 White participants

(68% female), with a mean age of 19.5 years (SD " 1.94). The

sample had a mean college class standing of 2.19 years (SD " 1.18

years). There were 26 Latino/a participants and 26 White partici-

pants assigned to the cross-group condition; they were paired into

26 cross-group dyads (69% female dyads). There were 54 White

and 38 Latino/a participants assigned a same-group partner, for a

total of 27 White same-group dyads (67% female) and 19 Latino/a

same-group dyads (84% female).

Experimenters. The dyads were randomly assigned to 1 of 11

experimenters who represented a diverse set of ethnic groups: 2

African Americans, 4 Asian Americans, 1 East Indian, 1 Latina,

and 3 Whites. The assigned experimenter ran all three friendship

meetings for each dyad. Experimenters were uninformed of the

specific hypotheses of the study.

Friendship Meetings

The friendship meetings began within 2 weeks of the information

session. We scheduled the meetings for 1 hr at the same time of day

each week for 3 consecutive weeks to account for the effects of

cortisol’s diurnal cycle within participant pairs (Lovallo & Thomas,

2000) and to allow for comparisons of cortisol changes across friend-

ship meetings. Partners were asked not to contact each other during

the 2 weeks that separated their three friendship meetings.

During each friendship interaction, participants completed an

expanded version of the Fast Friends procedure (Aron et al., 1997)

that we adapted for multiple meetings as outlined below. Our

paradigm was modeled on the original Fast Friends procedure

(Aron et al., 1997) and the multisession Fast Friends procedures

used to study cross-group friendship described by Wright et al.

(2002). We designed the expanded version of the Fast Friends

procedure to mimic the escalations in self-disclosure of the orig-

inal Fast Friends procedure and, in the final meeting, incorporated

a superordinate goal. Participants completed brief filler question-

naires at the beginning and end of each friendship meeting.

First friendship meeting. Participants arrived at the lab expect-

ing to complete a series of friendship-building tasks with a same-sex

friendship partner, but they did not know anything else about their

partner (i.e., the ethnicity of the partner was unknown). The first

meeting followed the Fast Friends procedure as outlined by Aron et al.

(1997), in which partners took turns reading and answering 36 ques-

tions that progressively encouraged self-disclosure.

We use the term ethnicity in this article to distinguish between our Latino/a

and White participants (cf. United States Census Bureau, Population Division,

Social and Demographic Statistics, 2003), recognizing nevertheless that dif-

ferences between these groups may be characterized as ethnic, racial, cultural,

or any mixture of these (Betancourt & Lo´pez, 1993; Phinney, 1996). Specif-

ically, because we classified White participants of Hispanic origin to be

Latino/a, we can only claim to have established friendships across ethnic

boundaries as opposed to racial ones. Nevertheless, we keep the term RS-race

to describe the RS-race questionnaire (Mendoza-Denton et al., 2002) to main-

tain consistency with prior literature.

The choice of a Latino/a and White sample was due to recruitment

access to Latino/a and Chicano/a student organizations. The choice of

ethnic groups reflects the work of Stephan and Stephan (1984, 1985), who

initially identified intergroup anxiety among Latinos/as and Whites.

1083

CROSS-GROUP FRIENDSHIP

Second friendship meeting. The second meeting was identical

to the first in protocol but used 36 new questions.

Potentialques

tions were drawn from the Book of Questions (Stock, 1987) and

were rated by an independent pilot sample (N " 16) on disclosure

prompted by the question and degree of comfort answering the

question with a stranger. We wanted the new questions to mimic

the same increase in self-disclosure as the original Fast Friends

procedure, and so the second set of Fast Friends questions was

matched to the original set on levels of comfort and self-disclosure.

The 36 final questions were rated similarly (M " 15.44, SD "

7.85) to the original 36 Fast Friends questions (M" 15.49, SD "

8.43), F(1, 35) " 1.15, ns, and thus were expected to elicit similar

levels of disclosure and closeness as the original Fast Friends

procedure.

Third friendship meeting. For their final meeting, the partners

played Hasbro’s Jenga, a game in which players remove blocks

from a stacked tower with the goal of not pulling the block that

makes the entire tower fall. Participants were offered an extra

ticket in end-of-semester lotteries for each block the pair success-

fully pulled from the tower before it fell. We chose Jenga because

Wright and colleagues (1998, 1999, 2002) used it successfully and

found it to be an effective way to incorporate play and shared

leisure activities in the development of friendship (see also Floyd

& Parks, 1995; McElwain & Volling, 2005). We also created a

cooperative goal for the friendship dyads (Allport, 1954; Sherif et

al., 1961) by instructing partners to develop a collective strategy to

increase their chances of winning the final lotteries.

Daily Diaries

Participants completed daily diaries for 10 consecutive evenings

beginning the day after the final friendship meeting. At 6:00 each

evening, participants were emailed a link to their daily diary Web

survey. Participants were asked to complete the diary before they

went to sleep each night, but at the latest by 8:00 the following

morning. Three participants did not have Internet access in the

evening, so their regular experimenters conducted the diaries over

the telephone. Participants were contacted by phone after the final

diary for debriefing and compensation.

Materials

All self-report measures were assessed using a Web-based survey

programmed in Perl 5.8 by an independent programmer (Miller,

2006). The surveys interfaced with an Apache 1.3 Web server located

on a lab computer, and data were stored in a password-protected

PostgreSQL database. Collections of free salivary cortisol and

reaction-time data are described in greater detail below.

Background Questionnaire

Participants completed a background questionnaire during the

latter half of the information session. The background question-

naire was administered through a Web survey interface in individ-

ual cubicles and took approximately 30 min to complete. The

descriptive statistics for the background measures are presented in

Table 1, and zero-order correlations between background measures

are presented in Table 2, with values for Latino/a participants

presented along the upper diagonal and values for White partici-

pants presented along the lower diagonal.

RS-race. The RS-race scale (Mendoza-Denton et al., 2002)

assesses expectations of rejection on the basis of ethnicity across

situations in which discrimination is applicable and possible (such

as cross-group interactions). Respondents indicate on a 6-point

Likert scale how anxious they would feel and how much they

would expect race-based rejection (discrimination, exclusion, un-

favorable treatment) across a number of situations (e.g., being

called on in class, a job interview). Anxious expectations are

conceptualized as “hot” cognitions (Metcalfe & Mischel, 1999)

whereby affect amplifies the effects of a given cognition. Thus, the

expectation score is multiplied by the anxiety score within each

situation (Downey & Feldman, 1996; Mendoza-Denton et al.,

2002). Scores across all situations are averaged to arrive at an

RS-race score, in which higher scores indicate greater anxious

expectation of race-based rejection. In this study, we administered

a 6-item version of the RS-race questionnaire (# " .83) composed

of highly loading items from the original RS-race questionnaire

(see Mendoza-Denton et al., 2002). We administered the same

scale to Latino/a and White participants. Reliabilities by ethnic

group are listed in Table 1.

Implicit prejudice. Implicit prejudice was assessed with a Cau-

casian–Latino/a lexical IAT (Greenwald et al., 1998, 2003) created

using DirectRT v2002 software from Empirisoft (Jarvis, 2002). Par-

ticipants were asked to sort words into four categories (bad, good,

Caucasian surnames, Latino/a surnames) using two keys. The top 10

most common White and Latino/a surnames in the United States from

the 1990 Census (United States Census Bureau, Population Division,

1995) were used as the ethnicity-related stimuli.

We chose lexical

stimuli (surnames) instead of visual stimuli (e.g., pictures of morphed

White and Latino/a faces) because visual stimuli may not readily

differentiate Whites and Latinos/as, whereas the surnames belonging

A detailed protocol for the expanded Fast Friends procedure and the 36

new questions are available at the Relationships and Social Cognition Lab

website: http://rascl.berkeley.edu/tools/

The 10 most common Latino/a and White surnames from the 1990

United States census (United States Census Bureau, Population Division,

1995) are available for use as lexical IAT stimuli. The surnames are

available at the Relationships and Social Cognition Lab website: http://

rascl.berkeley.edu/tools/

Table 1

Descriptive Statistics for Background Measures

Measure

Latino/a White

M SD N # M SD N #

RS-race 7.22 6.60 64 0.85 3.33 2.46 78 0.79

IAT $0.195 0.594 54 0.91 0.743 0.432 68 0.86

Prior contact 4.70 1.86 63 0.92 3.63 1.52 78 0.86

RS-personal 9.71 3.94 63 0.65 9.40 4.02 79 0.73

Note. RS-race " race-based rejection sensitivity; RS-personal " interper-

sonal rejection sensitivity.

1084

PAGE-GOULD, MENDOZA-DENTON, AND TROPP

to each group are reasonably distinctive.

Stimuli for the good and bad

categories were taken from Greenwald et al. (1998).

On half of the trials, participants were told to classify Latino/a

and good words with the same key and Caucasian and bad words

with the other (i.e., Latino/a favoritism block of trials). On the

other half of trials, Caucasian and good were paired on one key,

and Latino/a and bad were paired on the other key (i.e., Caucasian

favoritism block of trials). It is assumed that differences in cate-

gorization speed between the Caucasian favoritism and Latino/a

favoritism blocks represent the relative degree to which partici-

pants associated the concepts of good and bad with Caucasians and

Latinos/as. The presentation order of the Latino/a favoritism and

Caucasian favoritism blocks were randomized across participants.

The IAT D criterion was calculated according to the revised

scoring algorithm and error and response-latency corrections de-

scribed by Greenwald et al. (2003). We did not record response

latencies on practice blocks (Greenwald et al., 1998), and so our

only deviation from the revised algorithm (Greenwald et al., 2003)

was that we calculated the IAT D by only averaging reaction times

from the experimental blocks. Positive IAT D scores indicate bias

in favor of Caucasians and negative IAT D scores indicate bias in

favor of Latinos/as. The software we used to record the IAT

discarded some participants’ data when exiting the program, yield-

ing the lower number of completed IAT scores presented in Table

1. IAT scores were reliable (# " .89).

Prior intergroup contact. We assessed prior intergroup con-

tact using a measure developed by Islam and Hewstone (1993). We

controlled for prior intergroup contact to test the effects of making

a new cross-group friend above and beyond prior contact effects

and because prior intergroup contact may moderate the effects of

new contact experiences (Pettigrew & Tropp, 2006). As described

in the introduction, partners’ prior intergroup contact was also used

as a predictor when testing Hypothesis 2. White participants com-

pleted the measure with regard to their previous experiences with

Latinos/as, and Latino/a participants completed the measure with

regard to their previous experiences with Caucasians. The measure

includes five questions asking how much contact the participants

had with members of the other ethnic group in settings such as

college and outgroup members’ homes. Participants indicated how

often they had engaged in each form of contact using a 7-point

Likert scale (1 " no contact, 7 " a great deal of contact; # " .90).

Interpersonal rejection sensitivity. General sensitivity to so-

cial rejection was assessed using the Interpersonal Rejection Sen-

sitivity Questionnaire (Downey & Feldman, 1996). We included

interpersonal rejection sensitivity as a covariate to isolate the

influence of race-based rejection concerns from more general,

social rejection concerns (cf. Major & O’Brien, 2005). The Inter-

personal Rejection Sensitivity Questionnaire is structured in the

same way as the RS-race scale, in which respondents indicate how

anxious they would feel and how much they would expect rejec-

tion across six scenarios in which rejection is possible (e.g., “You

ask someone in one of your classes to coffee,” “You ask your

girlfriend/boyfriend if she/he really loves you”). We averaged the

products of the anxiety and expectation questions for each scenario

to create a composite score ranging from 1 to 36, where higher

values indicate greater interpersonal rejection sensitivity (# "

.72).

Laboratory Measures

Cortisol reactivity. Measuring free salivary cortisol is a non-

invasive method of obtaining hormonal stress responses. Cortisol

is released into the saliva approximately 20 min following the

onset of a stressor and starts to withdraw approximately 20 min

after the removal of the stressor (Dickerson & Kemeny, 2004). In

the present study, participants salivated into 1.9ml IBL-Hamburg

SaliCap collection vials in front of each other at the beginning and

end of each friendship meeting, for a total of six saliva samples per

participant. However, 36 saliva samples were not analyzable be-

cause of blood in the sample (bleeding gums) or insufficient levels

of saliva. Saliva samples were stored in a frost-resistant lab freezer

chest until they could be transported to an independent lab for

assay analysis. Intra- and interassay coefficients of variation were

3.26 and 5.05, respectively.

There is wide variation in baseline levels of cortisol, both

between and within individuals over the course of the day, and

thus, only relative changes in cortisol from baseline are psycho-

logically meaningful correlates of anxiety (Lovallo & Thomas,

2000), with positive scores indicating activation of the

hypothalamic-pituitary-adrenal (HPA) axis and rate of decline in

this activation after the removal of the stressor indicating recovery.

Although our cortisol scores may be interpreted as either cortisol

reactivity or recovery because of the time course of cortisol col-

lection, we retain the term cortisol reactivity to maintain consis-

tency with the literature (Dickerson & Kemeny, 2004). In line with

this literature, we computed cortisol reactivity scores for each

meeting by subtracting postmeeting cortisol from baseline (pre-

meeting) cortisol. These reactivity scores across the three friend-

ship meetings served as our physiological correlates of anxiety. To

To ensure against systematic differences between the lexical stimuli

representing each ethnic category, we compared the lengths and phonemes

of surnames for each group. The mean letters per surname were not

significantly different between ethnicities (M

White

" 6.2 letters, SD

White

1.14; M

Latino

" 6.1 letters, SD

Latino

" 0.99), F(1,19) " 0.04, ns, and the

standard deviations in word length were not different between ethnicities,

Levene’s F(1,18) " 0.26, ns. There were 44 total syllables across the 20

Latino/a and White surnames used, and only three syllables were shared by

both groups (all three were the phoneme “-or”), whereas 10 syllables out

of 24 (42%) were shared among Latino/a surnames and 9 syllables out of

20 (45%) were shared among White surnames.

Table 2

Zero-Order Correlation Matrix of Background Measures

Measure

(White)

Measure (Latino/a)

RS-race IAT

Prior

contact

RS-

personal

RS-race — $0.22 $0.3

0.15

IAT 0.01 — 0.23 $0.19

Prior contact 0.13 $0.004 — 0.031

RS-personal 0.11 0.001 $0.04 —

Note. RS-race " race-based rejection sensitivity; IAT " Implicit Associ-

ation Test (Greenwald, Nosek, & Banaji, 2003); RS-personal " interper-

sonal rejection sensitivity.

p % .05.

1085

CROSS-GROUP FRIENDSHIP

help account for the diurnal cycle, we ran participant dyads at the

same time of day each time they met.

Closeness with friendship partner. As a manipulation check,

closeness with friendship partner was assessed at the end of each

meeting using the Inclusion of Other in Self scale (IOS; Aron,

Aron, & Smollan, 1992). The IOS scale is a pictorial 7-point Likert

scale, depicting two increasingly overlapping circles, each labeled

“Self” and “Partner,” with greater overlap between the circles

depicting greater closeness. We chose to exchange “Partner” for

the circle typically labeled “Other” to make the scale more relevant

to the study. Participants were asked to select the picture that “best

represents your relationship with your interaction partner,” where

higher values indicate greater closeness.

Daily Diary Measures

Each diary consisted of filler questions to maintain our cover

story, such as daily mood, somatic symptoms, health behaviors,

and whether any classes had been missed. Among these mood

questions were five items assessing anxious mood (calm, nervous,

tense, attentive, worried) on a 7-point Likert scale. Calm and

attentive were reverse coded—participants’ responses were sub-

tracted from 8—so that higher values represented greater anxious

mood (average # " .90). We only examined anxious mood on

days when participants experienced cross-group interactions

(97.2% of total days reported), as we had too few instances of days

in which no cross-group interactions were reported. Participants

also reported on the nature of their longest social interactions with

someone of the same ethnic group and someone of another ethnic

group each day. If participants had not engaged in a cross-group

social interaction longer than 10 min that day, they were asked to

leave the questions blank. Participants were asked who had initi-

ated the interaction (self, other, or both), and degree of self-

initiated cross-group interactions were compared across friendship

conditions over the diary period.

Results

To address the three hypotheses outlined in the introduction, we

conducted analyses using multilevel linear modeling (Bryk &

Raudenbush, 1992; Kenny, Kashy, & Bolger, 1998; Singer &

Willett, 2003) implemented using the mixed procedure in SAS as

outlined by Singer (1998) and Bolger and Zuckerman (1995). The

multilevel approach permits the simultaneous analysis of within-

person and between-person variation, so it can account for the

nesting of participants within friendship pairs and within-person

dependence of repeated measures (Bolger & Zuckerman, 1995;

Kenny et al., 1998; Singer & Willett, 2003). In line with nested

growth-curve analyses (Singer & Willett, 2003), we modeled

effects over the friendship meetings using an unstructured covari-

ance matrix to account for the aforeme ntioned dependencies

within the data. We covaried participants’ own prior intergroup

contact and Interpersonal Rejection Sensitivity Questionnaire

score in all analyses below. In addition, for analyses involving

cortisol reactivity, we included the time of day during which a

given dyad’s meetings took place as a covariate to further control

for the effects of the diurnal cycle of cortisol. Prior to conducting

the analyses, we centered all continuous predictors (Aiken & West,

1991), with the exception of RS-race and IAT scores, which we

standardized within each ethnic group (see the Preliminary Anal-

yses section). We conducted parallel analyses using RS-race and

implicit prejudice as predictors.

Preliminary Analyses

Preliminary analyses revealed that cortisol reactivity was not

significantly affected by participant sex, participant age, or exper-

imenter ethnicity, and so these variables were not included in

subsequent analyses. Latinos’/as’ IAT scores were reversed so that

positive IAT scores would represent greater relative in-group bias

for all participants. Latinos/as and Whites differed significantly in

their mean IAT D scores, F(1, 122) " 104.12, p % .0001, and had

significantly different variances, Levene’s F(1, 121) " 3.97, p "

.049. Similarly, Latinos/as and Whites had significantly different

mean RS-race scores, F(1, 142) " 23.40, p % .0001, and unequal

variances, Levene’s F(1, 141) " 17.38, p % .0001. Thus, we

standardized these variables within ethnicity in the analyses below

to allow for comparisons between the ethnicities.

Manipulation Check: Growth in Partner Closeness

Self-reported partner closeness (as measured by the IOS) was

modeled as a function of friendship condition, ethnicity, meeting

number, and either RS-race or IAT with the aforementioned co-

variates. Only meeting number predicted changes in partner close-

ness, so we tested the main effects model of growth in partner

closeness as a function of meeting number, retaining the previous

predictors as covariates. Consistent with our expectations for the

expanded Fast Friends procedure, there was a strong increase in

partner closeness over the three friendship meetings (M

first

" 1.87,

first

" 1.28; M

second

" 2.85, SD

second

" 1.30; M

third

" 3.27,

third

" 1.34), b " 0.70, F(1, 273) " 135.31, p % .0001.

Reduction of Anxiety in Intergroup Contexts Through

Cross-Group Friendship

We modeled change in cortisol reactivity as a function of

friendship condition (same group vs. cross group), meeting num-

ber (first, second, third), ethnicity (Latino/a, White), and RS-race

(continuous), controlling for the effects of the variables described

above. This four-way interaction was not significant, b " $0.59,

F % 1, p " .492. Therefore, we collapsed across ethnicity but

retained it as a covariate in the final model presented in Table 3.

This model revealed a significant three-way interaction of friend-

ship condition, meeting number, and RS-race, b " $1.18, F(1,

309) " 7.23, p " .008. This interaction remained significant when

we controlled for participants’ implicit prejudice, b " $1.60, F(1,

251) " 11.61, p " .001.

To better understand this pattern, we examined lower order

effects in the context of the significant higher order interaction

(Aiken & West, 1991). In the cross-group friendship condition, the

RS-Race ! Meeting Number interaction was significant, b "

$0.64, F(1, 307) " 3.92, p " .049. This interaction is plotted in

Figure 1. As expected, cortisol reactivity attenuated significantly

as a function of meeting number, b " $1.18, F(1, 307) " 5.74,

p " .017. Simple slope analyses within each cross-group friend-

ship meeting revealed that RS-race predicted significantly greater

cortisol reactivity in the first meeting, b " 1.61, F(1, 307) "

1086

PAGE-GOULD, MENDOZA-DENTON, AND TROPP

10.69, p " .001, whereas there were no differences in cortisol

reactivity as a function of RS-race by the third friendship meeting,

b " 0.34, F % 1, p " .427. Among participants lower in RS-race

in the cross-group condition, however, meeting number did not

significantly predict cortisol reactivity, b " 0.14, F % 1, p " .791.

The Meeting Number ! RS-Race interaction approached sig-

nificance in the same-group condition, b " 0.54, F(1, 307) " 3.30,

p " .070, but cortisol reactivity did not significantly change over

the same-group friendship meetings for people low in RS-race,

b " $0.39, F(1, 307) % 1, p " .350, or for those high in RS-race,

b " 0.70, F(1, 307) " 2.55, p " .111. Thus, we did not pursue

findings from the same-group condition further.

We conducted a parallel analysis for implicit prejudice. The

four-way interaction of friendship condition, meeting number,

implicit prejudice, and ethnicity was not significant, b " 1.18, F(1,

245) " 1.63, p " .204. Collapsing across and controlling for

ethnicity yielded a marginal three-way interaction of friendship

condition, meeting number, and implicit prejudice, b " $0.83,

F(1, 252) " 3.13, p " .078. The significance of this interaction

was similar when we controlled for RS-race, b " $0.84, F(1,

251) " 3.33, p " .069. Even though this interaction did not reach

conventional significance levels, we proceeded to examine our a

priori planned comparisons (cf. Tybout & Sternthal, 2001). As

expected, the Implicit Prejudice ! Meeting Number interaction

was not significant in the same-group condition, b " $0.026, F %

1, p " .932, but was significant in the cross-group condition, b "

$0.85, F(1, 252) " 5.88, p " .016. As illustrated in Figure 2,

results for the cross-group friendship condition show that the slope

for cortisol reactivity was significantly negative as a function of

meeting number among participants higher in implicit prejudice,

b " $1.68, F(1, 252) " 7.45, p " .007, whereas no significant

relationship was observed between cortisol reactivity and meeting

number among participants lower in implicit prejudice, b "

$0.25, F % 1, p " .668. Implicit prejudice marginally predicted

heightened cortisol reactivity during the first cross-group meeting,

b " 1.01, F(1, 252) " 2.88, p " .09, and was reversed but not

significant by the final meeting, b " $0.71, F(1, 252) " 2.10, p

" .149.

Moderating Effects of Partners’ Prior Intergroup Contact

To address our second hypothesis, we ran the same analyses as

above but included an additional predictor in the model that was

not included in the prior analyses: the amount of prior intergroup

contact that each participant’s partner reported. No effects were

found with partner’s prior contact for implicit prejudice, either in

the full model or only when examined in the cross-group condi-

Figure 1. Cortisol reactivity as a function of race-based rejection sensi-

tivity (RS-race) over three cross-group friendship meetings. Least squares

means for high and low RS-race are estimated at one standard deviation

above and below the means of RS-race within each ethnic group, respec-

tively.

Figure 2. Cortisol reactivity as a function of implicit prejudice over three

cross-group friendship meetings. Least squares means for high and low

implicit prejudice are estimated at one standard deviation above and below

the means of implicit prejudice within each ethnic group. IAT " Implicit

Association Test (Greenwald, Nosek, & Banaji, 2003).

Table 3

Model Estimates of Cortisol Reactivity for RS-Race and Implicit

Prejudice (IAT)

Effect

RS-race

estimate (SE)

IAT

estimate (SE)

Intercept $1.35

(.35)

$1.49

(.42)

Hour 0.36

(.11)

0.38

(.13)

Prior contact 0.04 (.14) $0.07 (.16)

RS-personal $.0007 (.06) 0.01 (.07)

Ethnicity 0.47 (.47) 0.52 (.56)

RS-race/IAT 0.40 (.33) 0.22 (.38)

Friendship condition $.42 (.49) $.03 (.58)

(RS-Race/IAT) ! Friendship Condition 0.58 (.47) $0.08 (.57)

Meeting no. 0.16 (.30) 0.19 (.34)

(RS-Race/IAT) ! Meeting No. 0.54

†

(.30)

$0.03 (.31)

Friendship Condition ! Meeting No. $.65 (.49) $0.97

†

(.57)

(RS-Race/IAT) ! Friendship Condition

! Meeting No. $1.18

(.44)

$0.83

†

(.47)

Note. RS-race " race-based rejection sensitivity; RS-personal " interper-

sonal rejection sensitivity.

†

p % .10.

p % .05.

1087

CROSS-GROUP FRIENDSHIP

tion. Thus, we do not discuss effects for implicit prejudice further

in this section.

Analyses with RS-race, however, yielded a significant Friend-

ship Condition ! Meeting Number ! RS-Race ! Partner’s Prior

Contact interaction, b " $1.51, F(1, 294) " 9.36, p " .002.

Model estimates are shown in Table 4. As above, this interaction

was not moderated by participant ethnicity, b " $0.65, F % 1, p

" .446, yet remained robust when we controlled for participants’

implicit prejudice, b " $1.78, F(1, 238) " 11.63, p " .001.

Simple slope analyses revealed that the three-way interaction of

RS-race, meeting number, and partner’s prior contact was not

significant among participants in the same-group condition, b "

0.47, F(1, 294) " 1.44, p " .231, but was significant in the

cross-group condition, b " $1.04, F(1, 294) " 13.49, p " .0003.

As Figure 3 illustrates, the effect of meeting number was not

significant among participants lower in RS-race, regardless of

whether his/her partner had low contact, b " 0.26, F % 1, p "

.747, or high contact, b " 0.30, F % 1, p " .604.

An interesting pattern emerged for participants high in RS-race.

The slope for cortisol reactivity was significantly positive for

high-RS-race participants paired with an inexperienced partner,

b " 1.46, F(1, 294) " 4.89, p " .028. By contrast, this slope was

significantly negative for high-RS-race participants paired with an

experienced partner, b " $2.57, F(1, 294) " 25.86, p % .0001.

Among higher RS-race participants in particular, the effect of

partner prior contact was significant in the first friendship meeting,

b " 3.31, F(1, 294) " 32.9, p % .0001, but not significant by the

last friendship meeting, b " $0.79, F(1, 294) " 1.78, p " .238.

Effects Beyond the Friendship

To examine the effects of the friendship manipulation over time,

we assessed self-initiated cross-group interactions (vs. same-group

interactions) over the diary period and anxious mood on days when

cross-group interaction was reported. We analyzed diary reports

using an autoregressive covariance matrix to examine daily trends

while properly accounting for dependence within dyads and indi-

viduals across days (Bolger & Zuckerman, 1995).

Self-Initiated Cross-Group Interaction

We modeled self-initiated cross-group interactions on friendship

condition and either RS-race or implicit prejudice, including the

same covariates as above. There was no main effect of RS-race,

b " 0.09, F(1, 394) " 1.75, p " .193, nor was there an interaction

effect of RS-Race ! Friendship Condition, b " $0.05, F % 1, p "

.551, found for this dependent variable. For the model involving

implicit prejudice, however, we found a significant Friendship

Condition ! Implicit Prejudice interaction, b " $0.10, F(1,

394) " 4.64, p " .032. Model estimates are provided in Table 5.

The interaction remained significant when controlling for RS-race,

b " $0.10, F(1, 393) " 5.04, p " .025. As Figure 4 illustrates, the

effect of friendship condition was not significant among partici-

pants low in implicit prejudice, b " $0.06, F(1, 394) " .93, p "

.191, but was marginally significant among those high in implicit

prejudice, b " 0.12, F(1, 394) " 3.76, p " .059. The simple effect

of implicit prejudice on self-initiated interactions was not signifi-

cant either in the same-group condition, b " $0.05, F(1, 394) "

2.14, p " .191, nor in the cross-group condition, b " 0.05, F(1,

394) " 2.56, p " .146. To ensure that the findings were specific

to cross-group interactions, we ran a subsequent analysis that also

controlled for the frequency of self-initiated same-group interac-

Table 4

Model Estimates for Effects of Partners’ Prior Contact on

Cortisol Reactivity by RS-Race

Effects Estimate (SE)

Intercept $1.28

(.35)

Hour 0.36

(.11)

Prior contact 0.04 (.14)

RS-personal $0.02 (.74)

Ethnicity 0.47 (.48)

RS-race 0.44 (.38)

Friendship condition $0.56 (.49)

RS-Race ! Friendship Condition 0.69 (55)

Partner’s prior contact 0.30 (.30)

RS-Race ! Partner’s Prior Contact $.24 (.46)

Friendship Condition ! Partner’s Prior Contact 0.43 (.47)

RS-Race ! Friendship Condition ! Partner’s Prior

Contact 0.81 (.58)

Meeting no. 0.18 (.26)

RS-Race ! Meeting No. 0.61

†

(.33)

Friendship Condition ! Meeting No. $0.32 (.43)

RS-Race ! Friendship Condition ! Meeting No. $1.03

(.47)

Partner’s Prior Contact ! Meeting No. $.0005 (.27)

RS-Race ! Partner’s Prior Contact ! Meeting No. 0.47 (.40)

Friendship Condition ! Partner’s Prior Contact !

Meeting No. $0.98

(.42)

RS-Race ! Friendship Condition ! Partner’s Prior

Contact ! Meeting No. $1.51

(.49)

Note. RS-race " race-based rejection sensitivity; RS-personal " interper-

sonal rejection sensitivity.

†

p % .10.

p % .05.

Figure 3. Moderating effects of partners’ prior intergroup contact on

cortisol reactivity as a function of race-based rejection sensitivity (RS-race)

in the cross-group condition. Least squares means for both high and low

values of participants’ RS-race and partners’ prior intergroup contact are

estimated at a standard deviation above and below the means of each

predictor, respectively.

1088

PAGE-GOULD, MENDOZA-DENTON, AND TROPP

tions; this analysis continued to yield the Friendship Condition !

Implicit Prejudice interaction, now of marginal significance, b "

0.08, F(1, 393) " 3.64, p " .057.

Anxious Mood

Anxious mood was modeled as a function of friendship condi-

tion and either RS-race or implicit prejudice. The model involving

RS-race, despite not yielding a significant Friendship Condition !

RS-Race interaction, did reveal a significant main effect of RS-

race, b " 0.315, F(1, 470) " 8.84, p " .003, such that participants

higher in RS-race experienced greater anxious mood. The analysis

also yielded a compensatory main effect of friendship condition,

b " $0.41, F(1, 470) " 3.98, p " .0519, such that participants in

the cross-group condition felt less anxious over the period follow-

ing the friendship manipulation than did those in the same-group

condition. The main effect of RS-race remained significant when

we controlled for implicit prejudice, b " 0.29, F(1, 470) " 7.78,

p " .007, but the main effect of friendship condition became

marginally significant when we controlled for IAT, b " $0.42,

F(1, 470) " 3.50, p " .066. This pattern is illustrated in Figure 5.

No effects involving implicit prejudice were found for anxious

mood (all Fs % 1).

Discussion

The pattern of results reported here provide experimental evi-

dence for the benef its of cross-group friendship, particularly

among people who are most likely to experience anxiety in inter-

group contexts. Participants who were implicitly prejudiced or

concerned about outgroup rejection responded to the first cross-

group interaction with heightened cortisol reactivity. Despite this

initial display consisten t with intergroup anxiety (Step han &

Stephan, 1985), both participants high in implicit prejudice and

participants high in RS-race experienced significant decreases in

cortisol reactivity ove r three cross-group fri e ndship meetings.

These findings suggest that the attenuation of anxiety in intergroup

contexts that has been observed as a function of cross-group

friendship (e.g., Levin et al., 2003; Paolini, Hewstone, Cairns, &

Voci, 2004) can occur relatively early in the development of

cross-group friendship. Furthermore, over the week and a half after

the final cross-group friendship meeting, implicitly prejudiced

participants sought out more intergroup interactions, and partici-

pants felt less anxious in the diverse university environment that

constituted the setting for the study. These findings provide initial

experimental support for the process of intergroup anxiety reduc-

tion described by Paolini et al. (2006) as a function of cross-group

friendship.

Figure 5. Anxious mood on days with cross-group contact as a function

of race-based rejection sensitivity (RS-race) and friendship condition.

Least squares means for high and low RS-race are estimated at one

standard deviation above and below the mean of RS-race within each

ethnic group, respectively.

Table 5

Model Estimates for Self-Initiated Cross-Group Contact by

Implicit Prejudice

Effects Estimate (SE)

Intercept 0.21

(.03)

Prior contact 0.02 (.01)

RS-personal 0.005 (.005)

Ethnicity 0.09

†

(.05)

IAT $0.04 (.03)

Friendship condition 0.003 (.05)

IAT ! Friendship Condition 0.09

(.04)

Note. RS-personal " interpersonal rejection sensitivity; IAT " Implicit

Association Test (Greenwald, Nosek, & Banaji, 2003).

†

p % .10.

p % .05.

Figure 4. Effects of implicit prejudice and cross-group friendship on

initiation of cross-group interactions in daily life. Least squares means for

high and low implicit prejudice are estimated respectively at one standard

deviation above and below the mean of implicit prejudice within each

ethnic group. Self-initiated cross-group interaction was coded as a dummy

variable (where 1 indicated a self-initiated interaction), so the values for

initiation can be interpreted as the average likelihood of initiating cross-

group contact on a given day during the diary period. IAT " Implicit

Association Test (Greenwald, Nosek, & Banaji, 2003).

1089

CROSS-GROUP FRIENDSHIP

It is important to note that RS-race and implicit prejudice were

both related to cortisol increases in the first cross-group meeting.

Chronic activation of the HPA axis is associated with immuno-

suppression (Chrousos & Gold, 1992), heart disease (Whitworth,

Brown, Kelly, Williamson, 1995), Type II diabetes (Raikkonen,

Keltikangas-Jarvinen, Adlercreutz, & Hautenen, 1996), and cancer

(Wei et al., 2001). When paired with the inevitability of intergroup

interaction among members of minority groups (e.g., Hallinan &

Smith, 1985), this hormonal reaction to the first c ross-group

friendship meeting may have relevance to ethnic health disparities

in the United States across the aforementioned health conditions

(United States Department of Health and Human Services, 2000).

Our findings also suggest the possibility that cross-group friend-

ship may provide a coping mechanism for people who hold neg-

ative intergroup expectations, perhaps by providing repeated and

unambiguous disconfirmations of such expectations (McLaughlin-

Volpe, Mendoza-Denton, & Shelton, 2005; Mendoza-Denton et

al., 2006). An intriguing direction for future research is to directly

investigate the health implications of reductions in intergroup

anxiety as a function of cross-group friendship.

Lack of Differences as a Function of Ethnicity

We began this investigation with an expectation that ethnicity

would moderate the effects of RS-race and implicit prejudice. This

was based on the notion that concerns about being targets of

prejudice would be more relevant to minority group members and

that its converse, prejudice, would be more relevant to majority

group members. Although there were ethnic differences in the

means and variances of RS-race and implicit prejudice, we con-

sistently found that individual variation on these measures accom-

panied variations in anxiety in intergroup contexts similarly for

both Latino/a and White participants. This is consistent with

Stephan and Stephan’s (1985) view of prejudice and rejection

concerns as being applicable antecedents of intergroup anxiety for

all groups.

How can we reconcile the fact that the RS-race scale has been

previously found to be uniquely predictive among minority stu-

dents? One reason may be that prior research with RS-race has

largely focused on institutional outcomes (e.g., sense of belonging)

in universities where minority students have been traditionally

marginalized (e.g., Aronson & Inzlicht, 2004; Mendoza-Denton et

al., 2002; Mendoza-Denton et al., in press). White students have

less reason to question their belonging and acceptance in such

institutions because of their race, reducing the applicability of

institutional outcomes for this group. On the other hand, intergroup

anxiety seems applicable to both majority and minority group

members (Stephan & Stephan, 1985). Thus, even with the re-

stricted range of RS-race scores among White students that we

observed here and in prior studies, the applicability of anxiety in

intergroup social contexts for both groups—particularly in a di-

verse campus such as UC Berkeley—may render such variability

meaningful for Whites.

With respect to this last point, it is important to note that no

ethnic group constitutes a majority on the UC Berkeley campus,

although Asian American students are the plurality in this diverse

context. It is possible that our Latino/a and White samples re-

sponded similarly because they both represented members of nu-

merical minority groups within the particular institutional context

of the study. We interpret the lack of differences as a function of

ethnic group with caution given the nonrepresentativeness of the

setting and the moderate size of the sample in this study.

It is also worth noting that the psychological causes of RS-race

or implicit prejudice may be qualitatively different for majority

and minority group members. For a minority group member,

concerns about being rejected on the basis of race often center

around concerns about others’ perceptions of one’s belonging in a

particular milieu (Mendoza-Denton et al., 2002). Among majority

group members, by contrast, race-based rejection concerns may

center more around assumptions about one’s own prejudice (Goff

et al., 2008). Similarly, being prejudiced against a minority group

may not be associated with the same psychological motivations as

being prejudiced against a majority group, which has at times been

conceptualized as a self-protective reaction to stigma (Worrell,

Cross, & Vandiver, 2001). Although these differences may not be

differentially predictive for how much stress one experiences dur-

ing intergroup interaction, as suggested by the current findings,

other outcomes related to RS-race and implicit prejudice may

indeed be moderated by ethnicity.

Moderating Effects of Partners’ Prior Intergroup Contact

As Figure 3 illustrates, the effect of partner’s prior contact in the

first session was positive for cortisol reactivity among participants

high in RS-race, such that participants evidenced greater hormonal

stress responses the more contact their partners reported. This

finding seems to run counter to the intuition that participants with

greater intergroup contact should behave comfortably during in-

tergroup interactions and engender a positive atmosphere.

The literature on intergroup interaction suggests two possible

explanations for this finding. The first possibility is based on

research suggesting that high investment in the positivity of inter-

group interaction can lead people to “choke” under the pressure

and appear less positive and responsive in intergroup encounters

(Vorauer & Turpie, 2004). It is possible that high-contact part-

ners—who are presumably more motivated to build future inter-

group ties (Emerson et al., 2002)—might be especially likely to be

concerned about the impression they are making on outgroup

partners. For participants who were high in RS-race, less positivity

from an outgroup partner could be perceived as a rejection cue,

which in turn would trigger intense reactions (Mendoza-Denton et

al., 2002) that might be reflected in cortisol reactivity. The second

possibility is based on recent research on counterstereotypical

exemplars showing that ratings of an interaction partners’ atypi-

cality predicted physiological patterns consistent with threat (cf.

Blascovich, Mendes, Tomaka, Salomon, & Seery, 2003; Mendes,

Blascovich, Hunter, Lickel, & Jost, 2007). Assuming that partic-

ipants high in RS-race anxiously expect negativity in novel inter-

group settings (Mendoza-Denton et al., 2002), high–contact part-

ners may seem especially schema inconsistent and thus engender

threat among this group. Future research could collect video data

of the meetings to disentangle whether partners’ intergroup contact

was related to RS-race because the experienced partners behaved

negatively or with surprising aptitude.

Nonetheless, the effects of partners’ prior intergroup experience

seemed to catch up with the pair over time and exert the opposite

influence, a pattern suggested by Shelton, Richeson, Salvatore, and

Trawalter (2005, p. 401) when discussing ironic effects of preju-

1090

PAGE-GOULD, MENDOZA-DENTON, AND TROPP

dice. Thus, partners’ prior contact may have eventually had the

effect of making participants who were high in RS-race feel at

ease. However, this trend is difficult to interpret in the absence of

HPA activation.

Another question to consider is why might we have observed

partner effects for RS-race and not for implicit prejudice? The

findings for RS-race are consistent with the notion that being the

potential target of rejection sensitizes individuals to cues from

others, as research on interpersonal rejection sensitivity has shown

(e.g., Ayduk, May, Downey, & Higgins, 2003). Prejudiced atti-

tudes, however, may be applied to others, regardless of their

behavior or disposition; as such, people high in implicit prejudice

may be less likely to focus attentional resources on individuating

information about their partners and thus miss cues associated with

prior contact.

Effects Beyond the Friendship

A recent review of intergroup anxiety research (Paolini et al.,

2006) contends that if people who normally experience intergroup

anxiety have repeated positive intergroup contact, then chronic

experiences of intergroup anxiety may be attenuated. Though the

present study did not assess intergroup anxiety per se, our findings

offer initial support for the dynamic process of intergroup anxiety

change described by Paolini et al. (2006). More specifically, Pa-

olini et al. (2006) distinguished between episodic intergroup anx-

iety, which is anxiety experienced during any given intergroup

interaction, and chronic intergroup anxiety, which is a generalized

anxiety about intergroup interactions. According to these authors,

experiences of episodic intergroup anxiety feed chronic intergroup

anxiety; in turn, chronically feeling anxious about outgroup mem-

bers feeds anxiety experienced in future intergroup episodes. In the

present research, we observed reductions in anxiety during each of

three cross-group meetings, which would be consistent with re-

ductions in episodic intergroup anxiety. We further observed re-

ductions in anxious mood within a diverse social environment,

which can be roughly approximated to chronic intergroup anxiety.

Consistent with this interpretation, it seems that the develop-

ment of a new cross-group friendship at least temporarily in-

creased interest in intergroup interactions among implicitly preju-

diced participants. The increases in intergroup initiation among

implicitly prejudiced participants who made a cross-group friend

add to past findings regarding the role that intergroup anxiety can

play in avoidance of outgroup members (Islam & Hewstone, 1993;

Paolini, Hewstone, Cairns, & Voci, 2004). It is possible that

increased self-initiation of intergroup contact represents an early

phase of attitude change among implicitly prejudiced individuals,

whereby the development of closeness with an outgroup member

disconfirmed negative expectations and piqued curiosity about

people from other groups. Further, initiating more intergroup in-

teractions does not guarantee future friendships, as the initiation

finding could also represent an information-gathering phase. Here,

it is important to note that we did not test for such effects longer

than 10 days following the final friendship meeting, nor beyond

the particular institutional context of the study. Thus, a necessary

direction for future research is to chart the trajectory of behavioral,

emotional, and attitudinal change that may occur over an extended

period following the development of cross-group friendship.

We tested for effects of both RS-race and implicit prejudice on

anxious mood and cross-group initiation but observed that the

scales predicted intergroup outcomes in different domains of daily

life. Prejudice has long been related to avoidance of intergroup

contact (Allport, 1954; Paolini, Hewstone, Cairns, & Voci, 2004),

which might explain why implicit prejudice related specifically to

initiation of intergroup interaction following the formation of a

new cross-group friendship. Nonetheless, the experimental manip-

ulation of cross-group friendship appeared to improve anxious

mood among all participants in this setting. Further research on the

mechanisms through which cross-group friendship improves inter-

group interactions will serve to elucidate the domains of intergroup

interaction that are most relevant to attitudes versus social rejec-

tion concerns.

Limitations and Future Directions

Although these results are encouraging with respect to the

power of close positive intergroup contact, several methodological

limitations are worth noting. Despite the strength of our manipu-

lation to foster closeness among those most reticent about inter-

group interaction (i.e., high RS-race, high implicit prejudice), it is

important to recognize that this was a relatively safe environment,

where both the laboratory setting and the relatively diverse campus

environment provided sanction by authority (Allport, 1954). In

addition, the participants involved in this study were attracted by

the opportunity to make a new friend and were therefore motivated

to expand themselves through the development of a new relation-

ship (Aron, Aron, & Norman, 2001; Wright et al., 2002). These

conditions may explain why partner closeness increased steadily in

both friendship conditions, unaffected by RS-race or implicit prej-

udice, despite the more nuanced patterns for cortisol reactivity.

Such trends suggest that, even though everyone experienced sub-

jective increases in closeness, the experience of closeness was not

equally stress free for all participants across all time points. Future

research should explore whether a friendship manipulation would

be comparably successful in nonoptimal conditions or between

groups with histories of intense and violent conflict (cf. Islam &

Hewstone, 1993; Paolini, Hewstone, Cairns, & Voci, 2004).

It should also be noted that when participants first met, they

knew they would have to interact with their partner twice more in

the near future. Thus, our participants may have been particularly

motivated to have a good interaction during the initial friendship

meeting. Although this methodological feature diverges from past

intergroup anxiety research, the first meeting may reflect some

aspects of real-life cross-group interactions, especially between

acquaintances, where the possibility exists of meeting the outgroup

member on another occasion (e.g., Towles-Schwen & Fazio,

2003).

In this study, we systematized the content of the cross-group

interactions such that participants’ opportunity to discuss

ethnicity-related issues were generally minimized. However, some

theorists have argued that group membership must be salient

during intergroup contact for positive outcomes to generalize from

experiences with individual outgroup members to the outgroup as

a whole (see Brown & Hewstone, 2005; Hewstone & Brown,

1986). On the other hand, a cross-group friendship may provide a

safe environment in which to discuss difficult group-based issues

with an outgroup member, as the intergroup closeness existing

1091

CROSS-GROUP FRIENDSHIP

between friends implies a certain degree of trust and confidence

that the discussion can take place without discrimination (see

Tropp, in press; Tropp & Bianchi, 2007). Relatedy, Wright et al.

(1998, 2005) found that discussing racial issues as a part of the

friendship-building process did not undermine the positive effects

of friendship, and they proposed that such discussions may in fact

be important for promoting generalization. On the other hand, it is

conceivable that hearing a friend say something discriminatory or

prejudiced would likely be more hurtful than experiencing dis-

crimination from a stranger. Thus, a remaining question of interest

is whether conversations about group processes, such as racism or

stigmatization, are more productive when discussed among cross-

group friends, or whether such difficult topics unravel advances

towards closeness (Pettigrew & Tropp, 2006; Wright et al., 2002).

The findings reported here also have implications for research

surrounding t he common ingroup identity model (Gaertner,

Dovidio, Nier, Ward, & Banker, 1999). This model suggests that

viewing outgroup members as part of a common ingroup fosters

benefits, such as reduced intergroup bias (Gaertner, Rust, Dovidio,

Bachman, & Anastasio, 1994) and greater institutional belonging

and commitment among minority group members (Dovidio, Gaert-

ner, Flores Niemann, & Snider, 2001). Cross-group closeness may

facilitate this process, as the self-expansion theory of closeness

predicts that both ingroups and the group memberships of close

cross-group friends become included in the self-concept (Aron et

al., 2004). Thus, cross-group friendship may be a particularly

effective means to open the door to inclusive ingroup representa-

tions.

Conclusions

When our research is considered together with previous exam-

inations of cross-group friendship on intergroup anxiety (i.e.,

Levin et al., 2003; Paolini, Hewstone, Cairns, & Voci, 2004;

Paolini et al., 2006; Wright et al., 2002; Wright & van der Zande,

1999), there seems to be a consistent pattern: The formation of

cross-group friendship leads to reduced anxiety in intergroup con-

texts. The findings presented here causally demonstrate that cross-

group friendship reduces anxiety in intergroup contexts among

people who are predisposed to experience such anxiety. By com-

bining several methods to assess anxiety, we convergently tested

the anxiety-reducing effects of cross-group friendship within one

study. The longitudinal elements of the design allowed us to

observe the intermediate effects of friendship development. Future

research on cross-group friendship can therefore explore more

detailed questions, such as the boundary conditions under which

intergroup conflict outweighs the potential of friendship and the

mechanisms through which friendship shapes affective experi-

ences with outgroup members. Altogether, this experiment dem-

onstrated substantial benefits of cross-group friendship formation

across multiple manifestations of anxiety in intergroup contexts.

References

Aiken, L. S., & West, S. G. (1991). Multiple regression: Testing and

interpreting interactions. Thousand Oaks, CA: Sage.

Allport, G. W. (1954). The nature of prejudice. Cambridge, MA: Addison

Wesley.

Aron, A., Aron. E. N., & Norman, C. (2001). The self expansion model of

motivation and cognition in close relationships and beyond. In M. Clark

& G. Fletcher (Eds.), Blackwell handbook of social psychology, Vol. 2:

Interpersonal processes (pp. 478 –501). Malden, MA: Blackwell Pub-

lishers.

Aron, A., Aron, E. N., & Smollan, D. (1992). Inclusion of Other in the Self

scale and the structure of interpersonal closeness. Journal of Personality

and Social Psychology, 63, 596 –612.

Aron, A., McLaughlin-Volpe, T., Mashek, D., Lewandowski, G., Wright,

S. C., & Aron, E. N. (2004). Including others in the self. European

Review of Social Psychology, 15, 101–132.

Aron, A., Melinat, E., Aron, E. N., Vallone, R., & Bator, R. (1997). The

experimental generation of interpersonal closeness: A procedure and

some preliminary findings. Personality and Social Psychology Bulletin,

23, 363–377.

Aronson, J., & Inzlicht, M. (2004). The ups and downs of attributional

ambiguity: Stereotype vulnerability and the academic self-knowledge of

African American college students. Psychological Science, 15, 829–

836.

Ayduk, O., May, D., Downey, G., & Higgins, T. (2003). Tactical differ-

ences in coping with rejection sensitivity: The role of prevention pride.

Personality and Social Psychology Bulletin, 29, 435–448.

Betancourt, H., & Lo´pez, S. R. (1993). The study of culture, ethnicity, and

race in American psychology. American Psychologist, 48, 629 – 637.

Blascovich, J., Mendes, W. B., Hunter, S. B., Lickel, B., & Kowai–Bell, N.

(2001). Perceiver threat in social interactions with stigmatized others.

Journal of Personality and Social Psychology, 80, 253–267.

Blascovich, J., Mendes, W. B., Tomaka, J., Salomon, K., & Seery, M.

(2003). The robust nature of the biopsychosocial model challenge and

threat: A reply to Wright and Kirby. Personality and Social Psychology

Review, 7, 234–243.

Bolger, N., & Zuckerman, A. (1995). A framework for studying person-

ality in the stress process. Journal of Personality and Social Psychology,

69, 890 –902.

Britt, T. W., Boniecki, K. A., Vescio, T. K., Biernat, M., & Brown, L. M.

(1996). Intergroup anxiety: A Person ! Situation approach. Personality

and Social Psychology Bulletin, 22, 1177–1188.

Brown, R., & Hewstone, M. (2005). An integrative theory of intergroup

contact. Advances in Experimental Social Psychology, 37, 255–343.

Bryk, A. S., & Raudenbush, S. W. (1992). Hierarchical linear models:

Applications and data analysis methods. Newbury Park, CA: Sage.

Chrousos, G., & Gold, P. (1992). The concepts of stress and stress system

disorders. Journal of the American Medical Association, 267, 1244 –

1252.

Dickerson, S. S., & Kemeny, M. E. (2004). Acute stressors and cortisol

responses: A theoretical integration and synthesis of laboratory research.

Psychological Bulletin, 130, 355–391.

Dovidio, J. F., Gaertner, S. L., Flores Niemann, Y., & Snider, K. (2001).

Racial, ethnic, and cultural differences in responding to distinctiveness

and discrimination on campus: Stigma and common group identity.

Journal of Social Issues, 57, 167–188.

Dovidio, J. F., Kawakami, K., & Gaertner, S. L. (2002). Implicit and

explicit prejudice and interracial interaction. Journal of Personality and

Social Psychology, 82, 62–68.

Downey, G., & Feldman, R. H. L. (1996). Implications of rejection

sensitivity for intimate relationships. Journal of Personality and Social

Psychology, 70, 1327–1343.

Emerson, M. O., Kimbro, R. T., & Yancey, G. (2002). Contact theory

extended: The effects of prior racial contact. Social Science Quarterly,

83, 745–761.

Fazio, R. H., & Olson, M. A. (2003). Implicit measures in social cognition

research: Their meaning and use. Annual Review of Psychology, 54,

297–327.

Floyd, K., & Parks, M. R. (1995). Manifesting closeness in the interactions

of peers: A look at siblings and friends. Communication Reports, 8,

69 –76.

1092

PAGE-GOULD, MENDOZA-DENTON, AND TROPP

Gaertner, S. L., & Dovidio, J. F. (2005). Understanding and addressing

contemporary racism: From aversive racism to the common ingroup

identity model. Journal of Social Issues, 61, 615– 639.

Gaertner, S. L., Dovidio, J. F., Nier, J. A., Ward, C. M., & Banker, B. S.

(1999). Across cultural divides: The value of a superordinate identity. In

D. Prentice & D. Miller (Eds.), Cultural divides: Understanding and

overcoming group conflict (pp. 173–212). New York: Russell Sage

Foundation.

Gaertner, S. L., Rust, M. C., Dovidio, J. F., Bachman, B. A., & Anastasio,

P. A. (1994). The contact hypothesis: The role of a common ingroup

identity on reducing intergoup bias. Small Groups Research, 25, 224 –

249.

Goff, P. A., Steele, C. M., & Davies, P. G. (2008). The space between us:

Stereotype threat and distance in interracial contexts. Journal of Per-

sonality and Social Psychology, 94, 91–107.

Greenwald, A. G., McGhee, D. E., & Schwartz, J. L. K. (1998). Measuring

individual differences in implicit cognition: The Implicit Associations

Test. Journal of Personality and Social Psychology, 74, 1464 –1480.

Greenwald, A. G., Nosek, B. A., & Banaji, M. R. (2003). Understanding

and using the Implicit Associations Test: An improved scoring algo-

rithm. Journal of Personality and Social Psychology, 85, 197–216.

Hallinan, M. T., & Smith, S. S. (1985). The effects of classroom racial

composition on students’ interracial friendliness. Social Psychology

Quarterly, 48, 3–16.

Hewstone, M., & Brown, R. (1986). Contact is not enough: An intergroup

perspective on the contact hypothesis. In M. Hewstone & R. Brown

(Eds.). Contact and conflict in intergroup encounters (pp. 1– 44). Cam-

bridge, MA: Basil Blackwell.

Inzlicht, M., & Ben-Zeev, T. (2000). A threatening intellectual environ-

ment: Why females are susceptible to experiencing problem-solving

deficits in the presence of males. Psychological Science, 11, 365–371.

Islam, M. R., & Hewstone, M. (1993). Dimensions of contact as predictors

of intergroup anxiety, perceived outgroup variability, and outgroup

attitude: An integrative model. Personality and Social Psychology Bul-

letin, 19, 700–710.

Jarvis, W. B. G. (2002). DirectRT (Version 2002) [Computer software].

New York: Empirisoft.

Kenny, D. A., Kashy, D. A., & Bolger, N. (1998). Data analysis in social

psychology. In D. Gilbert, S. Fiske, & G. Lindzey (Eds.), The handbook

of social psychology (Vol. 1, 4th ed., pp. 233–265). Boston, MA:

McGraw-Hill.

Levin, S., van Laar, C. Y., & Sidanius, J. H. (2003). The effects of ingroup

and outgroup friendships on ethnic attitudes in college: A longitudinal

study. Group Processes and Intergroup Relations, 6, 76 –92.

Lovallo, W. R., & Thomas, T. L. (2000). Stress hormones in psychophys-

iological research: Emotional, behavioral, and cognitive implications. In

J. T. Cacioppo, L. G. Tassinary, & G. Berntson (Eds.), Handbook of

psychophysiology (pp. 342–367). New York: Cambridge University

Press.

Major, B., & O’Brien, L. T. (2005). The social psychology of stigma.

Annual Review of Psychology, 56, 393–421.

McElwain, N. L., & Volling, B. L. (2005). Preschool children’s interac-

tions with friends and older siblings: Relationship specificity and joint

contributions to problem behavior. Journal of Family Psychology, 19,

486 – 496.

McLaughlin-Volpe, T., Mendoza-Denton, R., & Shelton, N. (2005). In-

cluding out-group others in the self: Implications for coping with race-

based rejection and alienation among minority students. In G. Downey,

J. Eccles, & C. Chatman, (Eds.). Navigating the future: Social identity,

coping and life tasks (pp. 191–209). New York: Russell Sage Founda-

tion.

Mendes, W. B., Blascovich, J., Hunter, S., Lickel, B., & Jost, J. (2007).

Threatened by the unexpected: Physiological responses during social

interactions with expectancy-violating partners. Journal of Personality

and Social Psychology, 92, 698 –716.

Mendes, W. B., Blascovich, J., Lickel, B., & Hunter, S. (2002). Challenge

and threat during social interactions with White and Black men. Per-

sonality and Social Psychology Bulletin, 28, 939–952.

Mendes, W. B., Gray, H., Mendoza-Denton, R., Major, B., & Epel, E.

(2007). Why egalitarianism might be good for your health: Physiological

thriving during stressful intergroup encounters. Psychological Science,

18, 991–998.

Mendoza-Denton, R., Downey, G., Purdie, V. J., Davis, A., & Pietrzak, J.

(2002). Sensitivity to status-based rejection: Implications for African

American students’ college experience. Journal of Personality and So-

cial Psychology, 83, 896–918.

Mendoza-Denton, R., Page-Gould, E., & Pietrzak, J. (2006). Mechanisms

for coping with status-based rejection expectations. In S. Levin and C.

Van Laar (Eds.), Stigma and group inequality: Social psychological

perspectives (pp. 151–170). Mahwah, NJ: Erlbaum.

Mendoza-Denton, R., Pietrzak, J., & Downey, G. (in press). Distinguishing

institutional identification from academic goal pursuit: Interactive ef-

fects of ethnic identification and race-based rejection sensitivity. Journal

of Personality and Social Psychology.

Metcalfe, J., & Mischel, W. (1999). A hot/cool-system analysis of delay of

gratification: Dynamics of willpower. Psychological Review, 106, 3–19.

Miller, I. D. (2006). Implementing a psychological Web survey using open

source software. Retrieved January 1, 2008, from: http://iandennismill-

er.com/media/pdf/oss_survey(2006).pdf

Mischel, W. (1977). The interactions of person and situation. In D. Mag-

nusson & N. S. Endler (Eds.), Personality at the cross-roads: Current

issues in interactional psychology (pp. 333–352). Hillsdale, NJ: Erl-

baum.

Nosek, B. A. (2005). Moderators of the relationship between implicit and

explicit evaluation. Journal of Experimental Psychology: General, 134,

565–584.

Paolini, S., Hewstone, M., Cairns, E., & Voci, A. (2004). Effects of direct

and indirect cross-group friendships on judgments of Catholics and

Protestants in Northern Ireland: The mediation role of an anxiety-

reduction mechanism. Personality and Social Psychology Bulletin, 30,

770 –786.

Paolini, S., Hewstone, M., Rubin, M., & Pay, H. (2004). Increased group

dispersion after exposure to one deviant group member: Testing Ham-

burger’s model of member-to-group generalization. Journal of Experi-

mental Social Psychology, 40, 569 –585.

Paolini, S., Hewstone, M., Voci, A., Harwood, J., & Cairns, E. (2006).

Intergroup contact and the promotion of intergroup harmony: The influ-

ence of intergroup emotions. In R. Brown & D. Capozza (Eds.). Social

identities: Motivational, emotional, and cultural influences (pp. 209 –

238). Hove, East Sussex, United Kingdom: Psychology Press.

Pettigrew, T. F. (1998). Intergroup contact theory. Annual Review of

Psychology, 49, 65–85.

Pettigrew, T. F., & Tropp, L. R. (2005). Allport’s intergroup contact

hypothesis: Its history and influence. In J. F. Dovidio, P. Glick, and L.

Rudman (Eds.), On the nature of prejudice: Fifty years after Allport (pp.

262–277). Malden, MA: Blackwell Publishers.

Pettigrew, T. F., & Tropp, L. R. (2006). A meta-analytic test of intergroup

contact theory. Journal of Personality and Social Psychology, 90, 751–

783.

Phinney, J. S. (1996). When we talk about American ethnic groups, what

do we mean? American Psychologist, 51, 918–927.

Pin˜a, M., & Mendoza-Denton, R. (2003, April). Sensitivity to status-based

rejection among Mexican-American college students. Paper presented at

the Spartan Psychol ogical Ass ociation R esearch Co nference, Santa

Cruz, CA.

Plant, E. A., & Devine, P. G. (2003). The antecedents and implications of

1093

CROSS-GROUP FRIENDSHIP

interracial anxiety. Personality and Social Psychology Bulletin, 29,

790 – 801.

Raikkonen, K., Keltikangas-Jarvinen, L., Adlercreutz, H., & Hautenen, A.

(1996). Psychosocial stress and the insulin resistance syndrome. Metab-

olism: Clinical and Experimental, 45, 1533–1538.

Shelton, J. N., Richeson, J. A., & Salvatore, J. (2005). Expecting to be the

target of prejudice: Implications for interethnic interactions. Personality

and Social Psychology Bulletin, 31, 1189–1202.

Shelton, J. N., Richeson, J. A., Salvatore, J., & Trawalter, S. (2005). Ironic

effects of racial bias during interracial interactions. Psychological Sci-

ence, 16, 397–402.

Sherif, M., Harvey, O. J., White, J., Hood, W., & Sherif, C. W. (1961).

Intergroup conflict and cooperation: The robbers cave experiment.

Norman, OK: University Book Exchange.

Singer, J. D. (1998). Using SAS PROC MIXED to fit multilevel models,

hierarchical models, and individual growth models. Journal of Educa-

tion and Behavioral Statistics, 24, 323–355.

Singer, J. D., & Willett, J. B. (2003). Applied longitudinal data analysis:

Modeling change and event occurrence. New York: Oxford University

Press.

Snyder, M., & Ickes, W. (1985). Personality and social behavior. In G.

Lindzey & E. Aronson (Eds.), Handbook of social psychology (3rd ed.,

pp. 883–948). Reading, MA: Addison Wesley.

Stephan, W. G., & Stephan, C. W. (1984). The role of ignorance in

intergroup relations. In N. Miller & M. B. Brewer (Eds.), Groups in

contact: The psychology of desegregation (pp. 229 –257). New York:

Academic Press.

Stephan, W. G., & Stephan, C. W. (1985). Intergroup anxiety. Journal of

Social Issues, 41, 157–175.

Stephan, W. G., & Stephan, C. W. (1989). Antecedents of intergroup

anxiety in Asian-Americans and Hispanic- Americans. International

Journal of Intercultural Relations, 13, 203–219.

Stephan, W. G., & Stephan, C. W. (2000). An integrated threat theory of

prejudice. In S. Oskamp (Ed.), Reducing prejudice and discrimination:

The Claremont Symposium on Applied Social Psychology (pp. 23– 45).

Mahwah, NJ: Erlbaum.

Stock, G. (1987). The book of questions. New York: Workman Publishing

Company.

Towles-Schwen, T., & Fazio, R. H. (2003). Choosing social situations: The

relation between automatically activated racial attitudes and anticipated

comfort interacting with African Americans. Personality and Social

Psychology Bulletin, 29, 170–182.

Tropp, L. R. (2007). Perceived discrimination and interracial contact:

Predicting interracial closeness among Black and White Americans.

Social Psychology Quarterly, 70, 70 – 81.

Tropp, L. R. (in press). The role of trust in intergroup contact: Its signif-

icance and implications for improving relations between groups. In U.

Wagner, L. R. Tropp, G. Finchilescu, & C. Tredoux (Eds.), Improving

intergroup relations: Building on the legacy of Thomas F. Pettigrew.

Malden, MA: Blackwell Publishers.

Tropp, L. R., & Bianchi, R. A. (2007). Interpreting references to group

membership in context: Feelings about intergroup contact depending on

who says what to whom. European Journal of Social Psychology, 37,

153–170.

Tropp, L. R., & Pettigrew, T. F. (2005a). Differential relationships between

intergroup contact and affective and cognitive indicators of prejudice.

Personality and Social Psychology Bulletin, 31, 1145–1158.

Tropp, L. R., & Pettigrew, T. F. (2005b). Relationships between intergroup

contact and prejudice among minority and majority status groups. Psy-

chological Science, 16, 951–957.

Tybout, A., & Sternthal, B. (2001). Can I test for simple effects in the

presence of an insignificant interaction? Journal of Consumer Psychol-

ogy, 10, 5–10.

United States Census Bureau, Population Division. (1995). Frequently

occurring first names and surnames from the 1990 census. Retrieved

September 3, 2003, from http://www.census.gov/genealogy/names/

dist.all.last

United States Census Bureau, Population Division, Social and Demo-

graphic Statistics. (2003). U.S. Census Bureau guidance on the presen-

tation and comparison of race and Hispanic origin data. Retrieved April

8, 2007, from http://www.census.gov/population/www/socdemo/

compraceho.html

United States Department of Health and Human Services. (2000). Healthy

people 2010: Understanding and improving health (2nd ed.). Washing-

ton, DC: Department of Health and Human Services.

Vorauer, J. D., Hunter, A. J., Main, K. J., & Roy, S. A. (2000). Meta-

stereotype activation: Evidence from indirect measures for specific

evaluative concerns experienced by members of dominant groups in

intergroup interaction. Journal of Personality and Social Psychology,

78, 690 –707.

Vorauer, J. D., & Kumhyr, S. M. (2001). Is this about you or me? Self-

versus other-directed judgments and feelings in response to intergroup

interaction. Personality and Social Psychology Bulletin, 27, 706 –719.

Vorauer, J. D., & Sakamoto, Y. (2006). I thought we could be friends, but:

Sytematic miscommunication and defensive distancing as obstacles to

cross-group friendship formation. Psychological Science, 17, 326–331.

Vorauer, J. D., & Turpie, C. A. (2004). Disruptive effects of vigilance on

dominant group members’ treatment of outgroup members: Choking

versus shining under pressure. Journal of Personality and Social Psy-

chology, 87, 384–399.

Walton, G. M., & Cohen, G. L. (2007). A question of belonging: Race,

social fit, and achievement. Journal of Personality and Social Psychol-

ogy, 92, 82–96.

Wei, Z., Jin, F., Dunning, L. A., Shu, X., Qi, D., Wen, W., et al. (2001).

Epidemiological study of urinary 6&-hydroxycortisol to cortisol ratios

and breast cancer risk. Cancer Epidemiology Biomarkers & Prevention,

10, 237–242.

Whitworth, J. A., Brown, M. A., Kelly, J. J., & Williamson, P. M. (1995).

Mechanisms of cortisol-induced hypertension in humans. Steroids, 60,

76 – 80.

Worrell, F. C., Cross, W. E. J., & Vandiver, B. J. (2001). Nigrescence

theory: Current status and challenges for the future. Journal of Multi-

cultural Counseling and Development, 29, 201–213.

Wright, S. C., Aron, A., & Tropp, L. R. (2002). Including others (and

groups) in the self: Self-expansion and intergroup relations. In J. P.

Forgas & K. D. Williams (Eds.), The social self: Cognitive, interper-

sonal and intergroup perspectives (pp. 343–363). Philadelphia: Psychol-

ogy Press.

Wright, S. C., Brody, S. A., & Aron, A. (2005). Intergroup contact: Still

our best hope for reducing prejudice. In C. S. Crandall & M. Schaller

(Eds.), The social psychology of prejudice: Historical perspectives (pp.

115–142). Seattle, WA: Lewinian Press.

Wright, S. C., Ropp, S. A., & Tropp, L. R. (1998, August). Intergroup

contact and prejudice: Findings supporting the friendship hypothesis.

Symposium paper presented at the annual convention of the American

Psychological Association, San Francisco, CA.

Wright, S. C., & van der Zande, C. C. (1999, October). Bicultural friends:

When cross-group friendships cause improved intergroup attitudes. Pa-

per presented at the annual meeting of the Society for Experimental

Social Psychology, St. Louis, MO.

Received November 11, 2006

Revision received February 1, 2008

Accepted February 12, 2008 !

1094

PAGE-GOULD, MENDOZA-DENTON, AND TROPP