Journal of Personality and Social Psychology1994, Vol. 67, No. 6, 1012-1023

Copyright 1994 by the American Psychological Association, Inc.0022-3514/94/S3.00

Identity Negotiation in Roommate Relationships: The Self as Architectand Consequence of Social Reality

Shawn E. McNulty and William B. Swann, Jr.

The authors report two longitudinal studies of new college roommates (Ns = 69 and 95 pairs). Inboth studies, targets' initial self-views predicted changes in perceivers' appraisals of them, and per-ceivers' initial appraisals predicted changes in targets' self-views, although relatively few dyads dis-played both effects. The perceiver-driven and target-driven effects occurred when appraisals andself-views were negative as well as positive. Implications for self-verification theory and symbolicinteractionism are discussed, and a less restrictive model of how appraisals influence self-views isproposed.

Behavioral scientists have long viewed the self-concept as adynamic construct, inextricably linked to social life and socialrelationships. The identity negotiation framework offers one ofthe most comprehensive treatments of this interplay betweenself-concepts and the social world (Goffman, 1959; Swann,1987). The framework assumes that people initiate their in-teractions by (implicitly or explicitly) negotiating the identitiesthat they are to assume in the relationship. Such negotiationlays the groundwork for the smooth unfolding of the relation-ship and, ultimately, for the fulfillment of the goals that broughtthe parties together (Goffman, 1959; Stryker & Statham, 1985).

Various analyses have suggested that there are two faces to theidentity negotiation coin. On the one hand, the appraisals ofone person (arbitrarily dubbed the "perceiver") may influencethe self-concept of the other person (the "target"). At the sametime, the target's self-concept may influence the perceiver's ap-praisal of him or her. Systematic studies of these distinctiveforms of influence have spawned two independent lines of the-ory and research. We first consider evidence that the self is aconsequence of the appraisals of perceivers and then considerevidence that the self is the architect of the appraisals of others.

Shawn E. McNulty, Department of Sociology, Indiana University;William B. Swann, Jr., Department of Psychology, University of Texasat Austin.

This research was conducted while Shawn E. McNulty was a graduatestudent at the University of Texas. Data collection and the preparationof the manuscript were supported by research funds (Grant MH 37598)and a research scientist development award (Grant MH 00498) fromthe National Institute of Mental Health (NIMH) to William B. Swann,Jr. and by a postdoctoral fellowship in NIMH Training Program No.PHS-T32-MH14588-15 to Shawn E. McNulty during his tenure at In-diana University.

We are grateful to Sherry Franklin, Brian Giesler, John Sideris, andDenise Walton for their assistance in conducting this research; to DavidHeise, Dave Kenny, and Scott Long for statistical advice; and to GregHixon, Sheldon Stryker, and several anonymous reviewers for their help-ful comments on an earlier version of this article.

Correspondence concerning this article should be addressed toWilliam B. Swann, Jr., Department of Psychology, University ofTexas, Austin, Texas 78712. Electronic mail may be sent toswann@psyvax.psy.utexas.edu.

Self as Consequence of Social Reality: ReflectedAppraisal and Behavioral Confirmation Processes

Earlier in this century, the symbolic interactionists (e.g.,Cooley, 1902; Mead, 1934) proposed that people devise theirself-concepts by internalizing the appraisals of others (Stryker,1987). They also asserted that the link between perceivers' ap-praisals and targets' self-concepts is mediated by targets' per-ceptions of how they are viewed by others (i.e., their reflectedappraisals; Kinch, 1963). Variations of the reflected-appraisalhypothesis have endured over the years and currently representan important theme in many diverse areas, ranging from workon labeling and stigmatization (e.g., Becker, 1963) to investiga-tions of attachment (e.g., Bowlby, 1980; Sroufe & Fleeson,1986).

Although widely cited, the reflected-appraisal hypothesis hasreceived surprisingly little empirical support. One of the pri-mary sources of support has been cross-sectional research inwhich researchers have shown that perceivers' appraisals, re-flected appraisals, and self-concepts are related (e.g., Felson,1980,1985; Felson & Reed, 1986;Hoelter, 1984;Lundgren, Jer-gens, & Gibson, 1982; Schafer & Keith, 1985; for a review ofearlier research, see Shrauger & Schoeneman, 1979). A closelook at this research, however, indicates that targets' reflectedappraisals are more closely related to targets' self-concepts thanthey are to the actual appraisals of perceivers (Gecas, 1982;Shrauger & Schoeneman, 1979). By implying that reflected ap-praisals are merely a projection of targets' self-views, these dataundermine the possibility that such reflected appraisals medi-ate the link between perceivers' actual appraisals and the self-concepts of targets (or that such a link exists at all).

Undaunted, other advocates of the impact of appraisals onthe self have marshaled evidence that evaluations of perfor-mance or personality can alter the self-concepts of targets in lab-oratory settings (e.g., Jussim, Soffin, Brown, Ley, & Kohlhepp,1992; Shrauger & Lund, 1975; Shrauger & Schoeneman, 1979;Snyder & Swann, 1978). Critics have dismissed this research bynoting that the experimenter's presence and the demand char-acteristics of the setting may have compelled participants tochange their self-ratings.

Probably the strongest evidence that the self is influenced by

1012

IDENTITY NEGOTIATION AND SELF 1013

the appraisals of others comes from longitudinal field studies.Manis (1955), for example, found that college roommates' ap-praisals subsequently influenced targets' self-concepts. Unfor-tunately, because targets were influenced only when their room-mates viewed them more favorably than they viewed them-selves, variables such as the quality of the relationship (ratherthan appraisals per se) may have mediated Manis's findings.Nevertheless, Felson (1989) provided stronger evidence when heshowed that parents' appraisals predicted changes in children'sself-concepts. Similarly, Cole (1991) showed that the appraisalsof teachers and peers influenced the self-concepts of children.

Even the recent longitudinal field studies have not proven tobe above criticism, however. Neither the Cole (1991) nor theFelson (1989) studies established that reflected appraisals medi-ate the effects of appraisals on self-views, nor did they show thatappraisals influence the self-concepts of adults. This latter issueis particularly troublesome in light of recent contentions thatadults are impervious to the influence of reflected appraisals(e.g., Kenny & DePaulo, 1993). In addition, Jussim (1991)noted that evidence of appraisal effects on the self-views of chil-dren is clouded by the potential impact of objective accuracy onthe perceptions of both perceivers and targets. Conceivably, theadult perceivers may have developed objectively accurate ap-praisals of targets early in the relationship, and targets eventu-ally caught on, thus creating the illusion that perceivers wereinfluencing targets.

Of course, reflected appraisal processes are not the onlymechanism through which perceivers' appraisals may influencethe self-concepts of targets. One alternative is that perceivers'appraisals may influence their behavior toward targets, andtargets may in turn behaviorally confirm the initial appraisalsof perceivers. Such behavioral confirmation processes (e.g.,Merton, 1948; Rosenthal, 1976; Rosenthal & Jacobson, 1968;Rosenthal & Rubin, 1978; Snyder, Tanke, & Berscheid, 1977)may work in conjunction with self-perception processes to pro-duce self-concept change. That is, to the extent that targets ob-serve their own behavior and use it as a basis for inferring theirself-concepts (e.g., Bern, 1972; Fazio, Effrein, & Falender, 1981;Kenny & DePaulo, 1993; Snyder & Swann, 1978), behavioralchange should foster self-concept change. Although this behav-ioral confirmation-self-perception process is theoretically via-ble, there is no evidence that it occurs in naturalistic settings.

In short, the lack of definitive evidence that the appraisals ofperceivers influence the self-concepts of targets in naturalisticsettings has inspired suggestions that this venerable hypothesisbe abandoned. One purpose of our research was to determine ifsuch calls for abandonment are warranted.

The Self as Architect of Social Reality:Self-Verification Processes

Self-verification theory (Swann, 1983, 1990) asserts that peo-ple want others to validate and confirm their self-concepts—even when those self-concepts are negative. Such self-verifica-tion strivings are presumably motivated by a desire for coher-ence in mental and social life (Festinger, 1957; Lecky, 1945; Sec-ord & Backman, 1965; Swann, 1983; Swann, Stein-Seroussi, &Giesler, 1992).

Although substantial evidence supports the contention that

people work to maintain their negative self-views as well as theirpositive self-views (for a review, see Swann, 1990), few studieshave directly tested the prediction that targets will behave so asto bring perceivers' appraisals into harmony with their self-views. One was an experiment by Swann and Ely (1984). Targetswho were either self-perceived introverts or extroverts in-teracted with perceivers who had expectations of the targets thatclashed with the targets' self-views. By the end of the interac-tion, most targets had succeeded in changing the perceivers'minds. Although this study represents a clear demonstrationof self-concepts influencing appraisals in the laboratory, a fieldstudy by Felson (1989) failed to find convincing evidence thatchildren's self-concepts predict changes in their parents' ap-praisals. Because this may have been due to the dearth of powerthat children have relative to their parents, a field study in whichtargets interact with perceivers of equal status is in order.

Do Perceivers and Targets Resolve DisagreementsThrough Compromise?

Will identity negotiation typically result in a compromise be-tween initially discrepant views, with individual dyads exhibit-ing both appraisal effects and self-verification effects? The normof reciprocity (Gouldner, 1960; Regan, 1971) would seem tosuggest so, because, for example, a perceiver who revises his orher initial impression of a target by bringing it closer to thetarget's original self-view could be seen as making a concessionthat should elicit a reciprocal concession from the target. Al-though intuitively appealing, this reciprocal-compromise hy-pothesis requires that perceivers and targets be aware of theirinfluence on one another. To the extent that this condition is notmet, identity negotiation may be a largely unilateral affair, withsome dyads exhibiting appraisal effects on self-views and othersexhibiting self-verification.

Study 1

Study 1 was designed to track the influence of perceivers andtargets on one another. We also sought to examine the role oftargets' actual traits in the identity negotiation process. As Jus-sim (1991) noted, apparent effects of perceivers' appraisals ontargets' self-views may sometimes be due to the unmeasuredinfluence of targets' actual traits. Similarly, an apparent self-verification effect could be due to the target's self-view beingmore accurate than the perceiver's appraisal and to the per-ceiver eventually becoming aware of this fact by observing ex-ternal signs such as grades or the responses of people who haveknown the target for a longer period of time. To control for theinfluence of participants' actual abilities on the perceptions ofperceivers and targets, we obtained measures of targets' actualabilities in Study 1. Finally, we examined the percentage of dy-ads in our sample that exhibited compromise as opposed tomore unilateral negotiation outcomes.

Method

Participants

Participants in this study were 94 same-sex pairs of college room-mates from the University of Texas who were recruited to take part in a

1014 SHAWN E. McNULTY AND WILLIAM B. SWANN, JR.

study of roommate relationships during the fall semester of 1990. Atleast 1 member of each pair was a student in an introductory psychologyclass. Participants who were enrolled in Introductory Psychology par-ticipated to fulfill a course requirement; all other participants were paid$ 10 on completion of the study. Of the initial 94 pairs, 19 pairs failed tocomplete at least one session and were dropped from the study. This left75 pairs in the sample. Finally, we deleted 6 of the remaining roommatepairs because they had lived together previously, leaving 69 pairs (20male pairs and 49 female pairs) in the sample.1

ProcedureParticipants completed questionnaires containing ratings of the abil-

ities, personality traits, and global worth of both themselves and theirroommate (as well as items pertaining to another investigation) duringthe second and twelfth weeks of class. After participants had completedthe questionnaire at Time 1, the experimenter emphasized the impor-tance of not discussing their responses with their roommates until thestudy was over. Between the second and twelfth weeks of class, eachparticipant met once individually with an experimenter for a video-taped interview. During this interview, objective measures of abilitieswere administered. Participants were fully debriefed at the end of thefinal (twelfth week) session.

Measures

Appraisals and self-views. Participants used 10-point, percentile-based scales to rate both themselves and their roommates relative toother college students their own age on four abilities that have beenshown to be highly important to the self-concepts of undergraduates(Pelham & Swann, 1989). The four ability items were "intellectual/ac-ademic ability," "social skills/social competence," "athletic ability," and"physical attractiveness." Participants used the same 10-point scales torate themselves and their roommates on the Big Five personality traits(McCrae & Costa, 1987): neuroticism, extraversion, openness to newexperiences, agreeableness, and conscientiousness. To facilitate partici-pants' understanding of the items, we substituted the terms outgoing forextroverted and easy to get along with for agreeable on the question-naire. Similarly, to avoid drawing undue attention to the neuroticismitem (the only socially undesirable trait in the survey), we reframed it asemotional stability.

Participants rated their own global worth (i.e., self-esteem) using Ro-senberg's (1965) Global Self-Esteem Scale and rated their roommate'sglobal worth using an adaptation of the Rosenberg scale created by re-wording the items from the scale to refer to roommates. Items from thisglobal worth scale included "I see my roommate as a person of worth,at least on an equal basis with others," and "All in all, I'm inclined tofeel that my roommate is a failure." Both the self-esteem and globalworth measures were completed on 5-point scales anchored withstrongly disagree and strongly agree, and both measures were internallyconsistent (Cronbach's a for self-esteem = .89 at both Time 1 (Tl) andTime 2 (T2); a for global worth = .83 at Tl and .82 at T2).

Objective measures. We used participants' total Standard Achieve-ment Test (SAT) scores as an indication of academic ability. Measuresof social skills and physical attractiveness consisted of the averaged rat-ings of 5 female undergraduate judges who viewed videotapes of eachparticipant being interviewed by an experimenter and rated the partici-pant's social competence and physical attractiveness on 10-point scalesidentical to those used for self-ratings. The intra-class correlations were.56 for social skills and .62 for physical attractiveness. The measure ofathletic ability was derived from self-reports of high school athletic ac-tivities and awards (specifically, the number of semesters they played ona school team and the number of athletic awards and scholarships theyreceived). These three items were standardized and averaged to form anathletic ability index (a = .59). Because of the difficulty of achieving

Table 1Appraisal Effects in Study 1: Time 2 Self- Views Regressed onTime 1 (Tl) Self-Views and Tl Appraisals

Traits

Academic ability8

Social skills3

Athletic ability3

Attractiveness3

Neuroticism3

Extraversion3

Openness"Agreeableness3

Conscientiousness3

Global worthb

Tl

Pfinal

.77**'

.62**"

.75**"

.74**". .71**"

.76**

.70**

.67**

.37**

Self-view

R2

change

k .60**' .44*** .70**• 5 9 * *• 4 9 * *

" .63**• . 5 1 * *' .45**

.17**.76*** .58***

Tl Appraisal

0final

- .00.16*.16*.13

-.07.14*.19*.12.23*.11

R2

change

.00

.02*

.02*

.01

.00

.02*

.04*

.02

.05*

.01

R2

model

.60

.46

.72

.60

.49

.65

.55

.47

.22

.59

Note. Each row represents a hierarchical regression model in whichTl self-view was entered on the first step, and Tl appraisal was enteredon the second step. The /3 reported is the standardized regression co-efficient for the final equation. The R2 change is the increment in R2 atthe step on which the variable was entered.37V=68. bN=61.*p<. 05, one-tailed. **p<. 01, one-tailed. ***/><. 001, one-tailed.

consensus on what would constitute an objective measure of personalitytraits or global worth, we did not attempt to assess targets' actual stand-ing on these traits.

The means, standard deviations, and zero-order correlations of ap-praisals, self-views, and objective measures for each trait can be foundin the Appendix.

Results

Identity Negotiation Effects

Were targets' self-views influenced by perceivers' appraisalsas the semester progressed? Hierarchical regression analyses in-dicated that they were. For each trait, we regressed T2 self-viewsfirst on Tl self-views and then on Tl appraisals. The results ofthese analyses, displayed in Table 1, revealed that perceivers'appraisals reliably predicted T2 self-ratings of social skills, ath-letic ability, extraversion, openness to new experiences, and con-scientiousness when Tl self-ratings were controlled. BecauseManis (1955) found that appraisal effects occurred only whenthe appraisals were more positive than the initial self-concepts,we conducted follow-up analyses to determine if our effectswere similarly limited to appraisals that flattered the target. Foreach trait for which an appraisal effect was found, we hierarchi-cally regressed T2 self-views on T1 self-views, T1 appraisals, adummy variable that took on a value of 1 when T1 appraisalswere more positive than Tl self-views and took on a value of 0otherwise, and the interaction of this dummy variable with Tlappraisals. If the relative positivity of perceivers' initial apprais-als were a moderating variable, then these interaction terms

1 Reliable gender effects were neither expected nor found in the pres-ent studies. Similarly, the identity of the experimenter had no effect onthe results of either Study 1 (experimenters were two female undergrad-uates) or Study 2 (experimenters were two male graduate students).

IDENTITY NEGOTIATION AND SELF 1015

Table 2Self- Verification Effects in Study 1: Time 2 AppraisalsRegressed on Time 1 (Tl) Appraisals and Tl Self- Views

Tl Appraisal Tl Self-view

Traits

Academic ability*Social skills0

Athletic ability""Attractiveness8

Neuroticism"Extraversionb

Openness*Agreeableness*Conscientiousness*Global worth'

0final

.49**'

.54**"

.46**"

.69**'

.33**

.58**'

.79**"

.43**"

R2

change

> .30***• 3 9 * * *

* .38***K .44***

.11**k .42***" .63***• .18***

.50*** .26***

.52*** .27***

final

.37***

.28**

.30**- .12

.01

.25**

.09- .09

.01- .04

R2

change

.13***

.07**

.07**

.02

.00

.06**

.01

.01

.00

.01

R2

model

.43

.46

.45

.46

.11

.48

.64

.19

.26

.28

Note. Each row represents a hierarchical regression model in whichTl appraisal was entered on the first step and Tl self-view was enteredon the second step. The 0 reported is the standardized regression co-efficient for the final equation. The R2 change is the increment in R2 atthe step on which the variable was entered.*A r=68. bAf = 67. c A f =66 .** p< .01, one-tailed. *** p < .001, one-tailed.

should reach significance (Cohen & Cohen, 1975). They did not(all ts < 1, ns), indicating that targets internalized perceivers'negative appraisals as well as their positive appraisals.

Next, to test the prediction derived from self-verification the-ory that targets' self-views would influence perceivers' apprais-als, we regressed T2 appraisals first on Tl appraisals and thenon Tl self-ratings. The results, shown in Table 2, revealed reli-able self-verification effects for academic ability, social skills,athletic ability, and extraversion. Analogous to the appraisaleffects, follow-up analyses that tested the interaction of Tl self-ratings and a dummy variable that indexed the relative posi-tivity of those self-ratings found no evidence that perceiverswere influenced only by targets' positive self-views (all ts < 1,ns).

Actual Abilities

To determine if the appraisal and self-verification effects wehad observed for ratings of ability could be accounted for bythe impact of targets' actual abilities, we repeated the analysesreported in Tables 1 and 2 for academic ability, social skills,athletic ability, and physical attractiveness with the objectivemeasure of each trait as an additional predictor. Actual abilitieswere entered on the next to last step (i.e., immediately beforeappraisals when testing for appraisal effects and immediatelybefore self-ratings when testing for self-verification effects). Theonly reliable effect of actual ability to emerge was for physicalattractiveness, where the objective measure of attractivenesspredicted changes in targets' self-ratings of attractiveness (/J fi-nal = .18, R2 change = .03, t(54) = 2.12, p < .05, one-tailed).More important, all of the previously reported identity negoti-ation effects remained statistically reliable when actual abilitywas controlled. The largest change was for the effect of apprais-als of social skills on self-perceived social skills, where the beta

weight increased from .16 to .20 when actual skill wascontrolled.

Compromise?

The norm of reciprocity suggests that the outcome of identitynegotiation should be compromise. That is, dyads engaged inidentity negotiation should exhibit both appraisal effects (i.e., asmaller difference between T2 self-rating and Tl appraisal thanbetween Tl self-rating and Tl appraisal) and self-verificationeffects (i.e., a smaller difference between T2 appraisal and Tlself-rating than between Tl appraisal and Tl self-rating). Arough sense of how many negotiations resulted in compromisecan be obtained by classifying each dyad as exhibiting either anappraisal effect alone, a self-verification effect alone, botheffects, or neither effect. Though purely descriptive, such a clas-sification would shed some additional light on the types of iden-tity negotiation found in this sample. Accordingly, we classifieddyads in terms of negotiation outcome for each trait for whichthe regression analyses revealed evidence of identitynegotiation.

Dyads were classified as exhibiting both appraisal effects andself-verification effects if the absolute value of the difference be-tween T1 appraisal and T2 self-rating was less than the absolutevalue of the difference between Tl appraisal and Tl self-rating(an appraisal effect) and if the absolute value of the differencebetween T1 self-rating and T2 appraisal was less than the abso-lute value of the difference between T1 self-rating and T1 ap-praisal (a self-verification effect). Dyads that exhibited only oneof the two types of identity negotiation effects were classifiedas either appraisal-effect dyads or self-verification-effect dyads.Finally, dyads that exhibited neither type of effect were dividedinto those for whom identity negotiation was not possible (be-cause T1 appraisal equaled T1 self-rating) and those for whomnegotiation was possible but did not occur.

The percentages of dyads that fall into each category for traitsfor which evidence of identity negotiation was present are dis-played in Table 3. Inspection of Table 3 indicates that althoughsome of the dyads that engaged in identity negotiation compro-mised, many more did not. Thus, the norm of reciprocity doesnot appear to have been a major determinant of the outcome ofidentity negotiation in this sample.

Discussion

In summary, the results of Study 1 provide some of the firstevidence that, within naturally occurring relationships betweenadults, perceivers' appraisals will alter targets' self-views, andtargets' self-views will influence perceivers' appraisals. More-over, follow-up analyses found no evidence that these effectswere simply due to positivity strivings on the part of targets or tothe influence of targets' actual traits. Finally, individual dyadsdisplayed a variety of identity negotiation outcomes, with thenorm of reciprocity having little impact on the type of identitynegotiation that took place.

Before proceeding to describe Study 2, a note on the use ofcausal language is in order. Although one must always be carefulabout making causal inferences from nonexperimental data,the panel design used in the present studies substantially re-

1016 SHAWN E. McNULTY AND WILLIAM B. SWANN, JR.

Table 3Percentage of Dyads Exhibiting Appraisal Effect Only, Self- Verification Effect Only,Both Effects, and Neither Effect (Study 1)

Traits

Academic abilitySocial skillsAthletic abilityExtraversionOpennessConscientiousness

NoTldifference

23.230.413.023.510.124.6

Neithereffect

30.423.233.319.137.733.3

Appraisaleffect

13.018.817.416.220.321.7

Self-verification

effect

23.220.320.326.517.48.7

Botheffects

10.17.2

15.914.714.511.6

Note. Numbers are the percentage of dyads in each category. Results are displayed only for those traitswhere identity negotiation effects were found in the regression analyses. Tl = Time 1.

duces the uncertainty surrounding the interpretation of regres-sion effects. In a cross-sectional study, for example, finding thatappraisals predict self-views would be ambiguous because of thepossibility of reciprocal effects or correlated residuals (i.e., theregression weight could be misleading because it is picking upthe causal effect of the criterion on the predictor as well as theeffect of the predictor on the criterion). Use of a two-wave paneldesign avoids this problem because a criterion that is measuredat T2 cannot possibly have a causal effect on a predictor mea-sured at T1. Similarly, a panel design makes the task of generat-ing alternative explanations more difficult. To explain the ap-parent effect of appraisals on self-views with an omitted vari-able, for example, would require positing a variable that notonly predicted Tl appraisals and T2 self-views but also pre-dicted the variance in T2 self-views that remains after Tl self-views are controlled.

Study 2

Given the provocative nature of the results of Study 1, we feltit was important to replicate the study's key findings. Study 2was also designed to determine whether the effects of appraisalson self-views noted in Study 1 occurred because targets con-sciously recognized and internalized their roommates' apprais-als, as implied by symbolic interactionism. We addressed thisissue in two ways. First, we measured reflected appraisals totest the traditional reflected-appraisal hypothesis. Second, wemeasured targets' perceptions of how knowledgeable theirroommates were about their standing on various traits. Thismeasure allowed us to test a slightly less restrictive model ofperceiver influence, in which targets make conscious decisionsabout which perceivers are likely to provide valuable feedbackon a given trait, and these decisions moderate the impact of per-ceivers on targets. To make room for the additional items anditems relevant to another investigation, we deleted the ratings ofthe Big Five personality traits.

Method

ParticipantsThe participants in this study were 105 same-sex pairs of college

roommates from the University of Texas who were recruited to partici-

pate in a study of roommate relationships during the fall semester of1991. At least 1 member of each pair was enrolled in Introductory Psy-chology. Participants enrolled in Introductory Psychology received re-search credit for participating; all others were paid $ 10 at the end of thestudy. Eight pairs were dropped from the sample because of lost data,and 2 pairs were dropped because they had lived together previously.This left 95 pairs in the sample-23 pairs of men and 72 pairs of women.The average age of the sample was 18.3 years.

Procedure

Participants completed self-ratings and roommate ratings of abilitiesand global worth, as well as items related to another investigation, dur-ing the second and twelfth weeks of class. In addition to these measures,the final questionnaire contained ratings of reflected appraisals andtargets' perceptions of perceivers' knowledgeability about targets' traits.The order of the self-view and reflected-appraisal measures was coun-terbalanced across participants in the final questionnaire. None of theresults to be reported were affected by the order manipulation; as a re-sult, it will not be discussed further. Participants were urged not to dis-cuss their responses with their roommates after the first session and weredebriefed at the end of the final session.

Measures

Appraisals and self-views. Participants completed ratings of them-selves and their roommates on the same ability and global worth mea-sures used in Study 1. Once again, the self-esteem and global worthmeasures proved highly reliable at both points of measurement (self-esteem as = .87 and .86; global worth as = .82 and .86).

Reflected appraisals and knowledgeability estimates. We measuredreflected appraisals by having targets indicate how much of each of theability-related traits (i.e., academic ability, social skills, athletic ability,and physical attractiveness) they thought their roommates believed thatthey possessed on 8-point Likert scales anchored by none at all and alarge amount. Targets completed a similar measure of reflected apprais-als of global worth by indicating the extent to which they thought theirroommates would agree that they were "a good and worthwhile person"on a 5-point scale.

Before completing each of these reflected-appraisal measures, targetsindicated on 8-point scales how knowledgeable they felt their room-mates were about their standing on each trait. The corresponding itemfor global worth differed slightly in that it asked to what extent targets'roommates were in a good position to judge their overall worth as aperson, and was completed on a 5-point scale.

The means, standard deviations, and zero-order correlations of ap-

IDENTITY NEGOTIATION AND SELF 1017

Table 4Appraisal Effects in Study 2: Time 2 Self- Views Regressed onTime 1 (Tl) Self- Views and Tl Appraisals

Tl Self-view Tl Appraisal

Traits finalR2

change finalR2

changeR2

model

Academic ability" .59*** .40*** .18* .03* .43Social skills'" .65*** .40*** -.06 .00 .40Athletic ability8 .83*** .66*** -.04 .01 .67Attractiveness" .71*** .49*** -.03 .00 .49Global worth0 .83*** .68*** -.17f .03f .71

Note. Each row represents a hierarchical regression model in whichTl self-view was entered on the first step and Tl appraisal was enteredon the second step. The /3 reported is the standardized regression co-efficient for the final equation. The R2 change is the increment in R2 atthe step on which the variable was entered." # = 9 4 . " # = 9 3 . c iV=91.*p<. 05, one-tailed. *** p < .001, one-tailed. -fp< -01, two-tailed.

praisals, self-views, reflected appraisals, and knowledgeability estimatesfor each trait can be found in the Appendix.

Results

Identity Negotiation Processes

To determine if the basic identity negotiation effects noted inStudy 1 were present in this sample, we performed hierarchicalregressions of T2 self-views on Tl self-views and Tl appraisalsand T2 appraisals on Tl appraisals and Tl self-views for eachtrait. The results for T2 self-views in Table 4 indicate that per-ceivers' appraisals influenced targets' self-perceived academicability, whereas the results for T2 appraisals in Table 5 revealthat targets' self-views influenced perceivers' appraisals of aca-demic ability and athletic ability.2 As in Study 1, we conductedfollow-up analyses in which we sought to determine if the ap-praisal and self-verification effects noted were driven solely by

Table 5Self- Verification Effects in Study 2: Time 2 AppraisalsRegressed on Time 1 (Tl) Appraisals and Tl Self- Views

Tl Appraisal Tl Self-view

Traits

Academic ability"Social skills"Athletic ability"Attractiveness"Global worthb

0final

.70**"

.68**"

.72**"

.73**'

.66**1

R2

change

* .53**** .51***k .66**** .51***, 43***

(3final

.15*

.11

.20***- .05- .01

R2

change

.02*

.01

.03***

.01

.00

R2

model

.55

.52

.69

.52

.43

Note. Each row represents a hierarchical regression model in whichTl appraisal was entered on the first step and Tl self-view was enteredon the second step. The (3 reported is the standardized regression co-efficient for the final equation. The R2 change is the increment in R2 atthe step on which the variable was entered.•AT=94. b iV=92.* p < .05, one-tailed. *** p < .001, one-tailed.

relatively positive appraisals and self-views. Once again, thiswas not the case (all interaction ft < 1.32, ns).

Compromise?

Once again we classified the dyads in our sample into thosethat exhibited appraisal effects only, self-verification effectsonly, both effects, or neither effect. The results of this classifi-cation are displayed in Table 6 for the two traits for which reli-able identity negotiation effects were detected in the regressionanalyses. As in Study 1, only a minority of the dyads that en-gaged in identity negotiation compromised.

Reflected Appraisals as Mediators

Did the effects of perceivers' appraisals on targets' self-viewsproceed through conscious internalization, or were more subtleprocesses, such as behavioral confirmation and self-perception,at work? Our first attempt to answer this question involved test-ing the mediational model proposed by symbolic interaction-ism. Before describing our test of the reflected-appraisal hy-pothesis, it is important to note that such a mediational modelcould be tested meaningfully only for ratings of academic abil-ity. This is because academic ability was the only trait in thepresent study to show a reliable effect of appraisals on self-views, and a statistically reliable effect of appraisals on self-views is a necessary precondition for testing any mediationalmodel (Judd & Kenny, 1981a, 1981b).

In addition to perceivers' appraisals predicting changes inself-views, the reflected-appraisal hypothesis requires that per-ceivers' appraisals predict reflected appraisals and that reflectedappraisals predict self-view change when perceivers' appraisalsare controlled (Judd & Kenny, 1981a, 1981b). We tested thefirst requirement, that actual appraisals predict reflected ap-praisals, by regressing T2 reflected appraisals of academic abil-ity first on Tl self-ratings of academic ability and then on Tlappraisals of academic ability. We entered self-ratings first be-cause of previous suggestions that reflected appraisals may benothing more than projections of self-views (Shrauger &Schoeneman, 1979). As shown in the highlighted portion of Ta-ble 7, perceivers' appraisals of academic ability did reliably pre-dict reflected appraisals of academic ability with initial self-views controlled. Appraisals also reliably predicted reflected ap-praisals for social skills and athletic ability.

The highlighted section of Table 8, however, reveals that whenactual appraisals of academic ability were controlled, reflectedappraisals did not reliably predict changes in self-rated aca-demic ability. Thus, reflected appraisals did not mediate theeffect of perceivers' appraisals on self-view change in thissample.

2 We also found an unexpected negative appraisal effect for globalworth (0 = - .17 , f[89] = -3.05, p < .01, two-tailed). The results offollow-up analyses on global worth suggested that this effect was drivenby self-esteem increases among targets who were initially appraised neg-atively by their roommates. The origin of this counterintuitive findingremains unclear. It should be noted that there was no sign of such aneffect in Study 1.

1018 SHAWN E. MCNULTY AND WILLIAM B. SWANN, JR.

Table 6Percentage of Dyads Exhibiting Appraisal Effect Only, Self-Verification Effect Only,Both Effects, and Neither Effect (Study 2)

Traits

Academic abilityAthletic ability

NoTldifference

28.427.4

Neithereffect

27.431.6

Appraisaleffect

16.813.7

Self-verificationeffect

16.817.9

Botheffects

10.59.5

Note. Numbers are the percentage of dyads in each category. Results are displayed only for those traitswhere identity negotiation effects were found in the regression analyses. Tl = Time 1.

Other Effects of Reflected Appraisals on Self- Views

Although the results of our analyses of academic ability rat-ings did not support the traditional mediational model of re-flected appraisals, other analyses suggest that reflected apprais-als do have a role to play in determining self-views. In particular,Table 8 indicates that reflected appraisals did reliably predictself-ratings of social skills, athletic ability, and physical attrac-tiveness when actual appraisals were controlled.3 What is some-what surprising about these effects is that for none of these threetraits was there any effect of actual appraisals on self-views, andthat, in the case of physical attractiveness, there was not even areliable effect of actual appraisals on reflected appraisals. Theseresults suggest that an expanded model of the genesis of re-flected appraisals and their role in determining self-views maybe needed. We return to this issue in the Discussion.

Perceived Knowledgeability

Another way of asking whether targets were aware of perceiv-ers' appraisals and consciously used them to modify their self-views is to determine if targets were more likely to be influencedby appraisals from roommates whom they believed to beknowledgeable about particular traits. To address this issue, foreach trait we regressed T2 self-views on T1 self-views, T1 ap-

Table 7Effect of Time 1 (Tl) Appraisals on Time 2 ReflectedAppraisals with Tl Self-Views Controlled (Study 2)

Traits

Academic ability1

Social skills"Athletic ability"Attractiveness'Global worth"

Tl

fina

.31**

.39**

.40**

.47**

.30**

Self-view Tl Appraisal

R2 0change final

* .12**'* .20**'* .25*** .24*** .09**

• . 1 7 *k .23**• . 2 2 ** .13' .04

R2

change

.03*

.05**

.04*

.02

.00

R2

model

.15

.25

.29

.26

.09

Note. Each row represents a hierarchical regression model in whichTl self-view was entered on the first step and Tl appraisal was enteredon the second step. The 0 reported is the standardized regression co-efficient for the final equation. The R2 change is the increment in R2 atthe step on which the variable was entered. The model relevant to themediational argument is in boldface type.a;V=94. bN=92.*p<.05,one-tailed. **p<.01,one-tailed. ***p< .001,one-tailed.

praisals, targets' ratings of perceivers' knowledgeability abouttheir standing on the trait, and the interaction of Tl appraisalsand knowledgeability estimates (in that order).

For two of the traits, the interaction of Tl appraisals andknowledgeability estimates emerged as a reliable predictor offinal self-views (athletic ability: 0 final = .18, R2 change = .03,/(90) = 2.94, p < .001, and physical attractiveness: /3 final = .26,R2 change = .06, ?(90) = 3.41, p < .001, one-tailed). In bothcases the form of the interaction was such that targets' self-viewsbecame more positive to the extent that they were evaluatedpositively by a credible roommate or were evaluated negativelyby a less credible roommate. Similarly, targets' self-views be-came more negative if they were evaluated negatively by a cred-ible roommate or if they were evaluated positively by a less cred-ible roommate.

Discussion

The data from Study 2 substantially replicated the identitynegotiation effects found in Study 1. As in the previous study,both appraisal effects and self-verification effects were found inthe regression analyses, and neither effect was limited to casesin which the driving force (appraisal or self-view) was positive.Also consistent with Study 1, only a small percentage of dyadsdisplayed both identity negotiation effects.

Contrary to the traditional form of the reflected-appraisal hy-pothesis, reflected appraisals did not mediate the effect of actualappraisals on self-view change. Reflected appraisals did, how-ever, have an impact on self-view change independent of per-ceivers' actual appraisals.

Finally, targets were more likely to be influenced by perceiv-ers' appraisals of their traits when they believed that the per-ceiver was knowledgeable about their standing on those traits.This finding supports Shrauger and Schoeneman's suggestionthat observer knowledgeability is an important and often over-looked variable in studies of the role of appraisals in self-viewchange (Shrauger & Schoeneman, 1979, p. 567; see also Rosen-berg, 1973).

3 To assure ourselves that the superiority of reflected appraisals toactual appraisals for predicting these self-views was not an artifact oftiming (actual appraisals were measured at T1; reflected appraisals weremeasured at T2), we reanalyzed the data with T2 appraisals and re-flected appraisals as the predictors. Again, reflected appraisals, but notactual appraisals, predicted changes in self-rated social skills, athleticability, and physical attractiveness.

IDENTITY NEGOTIATION AND SELF 1019

Table 8Effect of Time 2 (T2) Reflected Appraisals on T2 Self-Views With Time 1 (Tl) Self-Views andTl Appraisals Controlled (Study 2)

Traits

Academic ability'Social skills6

Athletic ability"Attractiveness8

Global worth'

Tl

final

.57***

.59***

.78***

.57***

.84***

Self-view

R2

change

.40***

.40***

.66***

.49***

.68***

Tl

0final

.16*-.10-.06-.06-.17t

Appraisal

R2

change

.03*

.00

.00

.00

.03t

T2 Reflectedappraisal

0 R2

final change

.08 .00

.16* .02*

.12* .02*

.28*** .06***-.00 .00

R2

model

.43

.42

.68

.55

.71

Note. Each row represents a hierarchical regression model in which Tl self-view was entered on the firststep, T1 appraisal was entered on the second step, and T2 reflected appraisal was entered on the third step.The 0 reported is the standardized regression coefficient for the final equation. The R2 change is the incre-ment in R2 at the step on which the variable was entered. The mediational model is in boldface type.a7V=94. bJV=93. C7V=91.* p < .05, one-tailed. *** p < .001, one-tailed, t P < 01, two-tailed.

General Discussion

Although much remains to be learned about the process andoutcome of identity negotiation, these initial studies have dem-onstrated that empirical research that focuses on naturally oc-curring relationships can lead to a productive merger of workon the social construction of the self and self-verification. Thepresent research has not only confirmed that both perceiver-driven effects and self-driven effects are at work in the everydaylife of college students, but it has also shed new light on eachphenomenon.

Appraisal Effects

The data on appraisal effects from these studies belie con-cerns that such effects might be limited to laboratory settings,to appraisals that flatter the recipient, or to the malleable self-views of children. Of course, skeptics could point out that themagnitude of our appraisal effects was small. In light of the so-cial norms against openly communicating appraisals and themuch larger knowledge base presumably supporting self-views,however, we believe that even tiny appraisal effects testify to theimportant impact of social feedback on self-views. This point iseven more telling when one reflects on the fact that the presentstudies observed the influence of only one of the many promi-nent individuals who compose participants' social worlds—adrop in the social psychological bucket, as it were.

Both our theoretical analysis and data suggest that the re-flected-appraisal hypothesis, as traditionally formulated, is toorestrictive to account for the full impact of social appraisals onthe self. One way to make the model less restrictive would beto broaden our understanding of what our reflected-appraisalmeasures are tapping. Kenny & DePaulo (1993), for example,have shown that people form relatively accurate impressions ofhow they are viewed by people in general, but their specific re-flected appraisals are less accurate. If people typically have well-formed impressions of how they are viewed in general (i.e., gen-eral reflected appraisals) but have only sketchy impressions of

how specific people view them (i.e., specific reflected apprais-als), then they may respond with a general reflected appraisalwhen asked for a specific reflected appraisal by a researcher. Ifso, then a measure of reflected appraisals may predict self-viewchange even when the appraisal it nominally reflects does not(as in Study 2), because it serves as a proxy for the collectiveappraisals of the wider social community.

Another way of easing the restrictions on the explanation ofappraisal effects is to consider mechanisms that do not involvethe conscious internalization of reflected appraisals. The behav-ioral confirmation-self-perception process is one such mecha-nism. Alternatively, identity negotiation may proceed throughsubtle statements and nonverbal cues that largely bypass con-scious awareness (e.g., Blumberg, 1972; Swann, Stein-Seroussi,&McNulty, 1992).

Self-Verification

Our data provide the first nonlaboratory evidence that peopleself-verify by bringing others to see them as they see themselves.Taken together with earlier evidence that people are less com-mitted to marital partners who appraise them incongruently(Swann, Hixon, & De La Ronde, 1992), these data provide evi-dence that people both seek and find self-verification in natu-rally occurring relationships. Moreover, the relative effect sizesacross Studies 1 and 2 indicate that self-verification is at leastan equal partner with appraisal effects in identity negotiationand may even be the stronger of the two processes. Although itis certainly too soon to conclude that self-verification processesare more powerful than appraisal effects in daily interactions,the analogous result has been demonstrated in the laboratory(Levesque & Kenny, 1994; Swann & Ely, 1984). In addition tothe relative size of the self-verification effects, one should alsonote that they, like the appraisal effects, were driven by negativeas well as positive beliefs. The evidence that people broughttheir relationship partners to recognize their negative self-viewsas well as their positive self-views casts doubt on claims that

1020 SHAWN E. McNULTY AND WILLIAM B. SWANN, JR.

people are exclusively motivated by a desire for positive evalua-tions (e.g., Jones, 1973).

Certainly we do not deny the existence of a powerful desirefor positivity; in fact, we have repeatedly encountered evidenceof this desire in our own laboratory (e.g., Hixon & Swann, 1993;Swann, Griffin, Predmore, & Gaines, 1987; Swann, Hixon,Stein-Seroussi, & Gilbert, 1990). We suggest instead that thedesire for positivity and the desire for self-verification are quitedistinct from one another (e.g., Swann, 1990; Swann &Shroeder, 1994) and that it is thus inappropriate to subsumeone under the other. For example, Steele, Spencer, and Lynch(1993) would suggest that our participants brought their room-mates to recognize their flaws in an effort to feel morally andadaptively adequate. Similarly, Baumeister (1993) would sug-gest that roommates sought self-verification of negative self-views in an effort to "protect" themselves against negative feed-back they might encounter if they were to seek overly favorableevaluations. Such formulations have the appeal of being able toexplain everything, including a preference for favorable evalua-tions, a preference for unfavorable evaluations, or no preferencewhatsoever. Yet these formulations are so broad that they arehardly distinguishable from reinforcement theory, which hy-pothesizes that people have a tendency to seek pleasure andavoid pain. Unfortunately, like reinforcement theory, these for-mulations are not falsifiable. We therefore remain convincedthat positivity and self-verification strivings are best understoodas being distinct.

Finally, we do not claim that self-verification effects are ubiq-uitous. Felson (1989), for example, concluded that children'sself-views had little or no impact on the appraisals of their par-ents. Two variables may have contributed to Felson's failure touncover evidence that children evoke self-verifying appraisalsfrom their parents. One is that the lower status of children un-dermined their ability to influence their parents' appraisals ofthem; our data indicate that perceived knowledgeability maybe an important moderator of identity negotiation processes.Alternatively, the self-views of children may be less certain thanthe firmly held self-views of adults and thus be less likely toinspire self-verification strivings (Swann, 1990). In either case,our findings suggest that adults are quite capable of obtainingself-verification from their relationship partners.

Summary and Conclusion

The portrait of the self that emerges from our data is one inwhich the self is both an architect and consequence of others'appraisals. More specifically, just as the self-concepts of our par-ticipants were shaped by the appraisals of others, they also in-fluenced those appraisals. Through this process, the self and thesurrounding social environment are constantly attuning them-selves to one another. Future research should consider whatvariables may lead a specific negotiation to be relatively target-driven or perceiver-driven and should also take a fresh look atthe types of mechanisms that may underlie the identity negoti-ation process. By attending to the power of both perceivers' ap-praisals and targets' self-views, such research may begin to laybare the intricate manner in which identity negotiation is woveninto the fabric of social relationships.

References

Baumeister, R. F. (1993). Understanding the inner nature of low self-esteem: Uncertain, fragile, protective, and conflicted. In Self-esteem:The puzzle of low self-regard (pp. 201-218). New York: Plenum.

Becker, H. W. (1963). Outsiders: Studies in the sociology of deviance.New York: Free Press.

Bern, D. J. (1972). Self-perception theory. Advances in ExperimentalSocial Psychology, 6, 2-62.

Blumberg, H. H. (1972). Communication of interpersonal evaluations.Journal of Personality and Social Psychology, 23, 157-162.

Bowlby, J. (1980). Attachment and loss: Loss (Vol. 3). New York: BasicBooks.

Cohen, J., & Cohen, P. (1975). Applied multiple regression/correlationanalysis for the behavioral sciences. Hillsdale, NJ: Erlbaum.

Cole, D. A. (1991). Change in self-perceived competence as a functionof peer and teacher evaluation. Developmental Psychology, 27, 682-688.

Cooley, C. H. (1902). Human nature and the social order. New York:Scribner's.

Fazio, R. H., Effrein, E. A., & Falender, V. J. (1981). Self-perceptionsfollowing social interaction. Journal of Personality and Social Psy-chology, 41, 232-242.

Felson, R. B. (1980). Communication barriers and the reflected ap-praisal process. Social Psychology Quarterly, 43, lli-l'i'h.

Felson, R. B. (1985). Reflected appraisal and the development of self.Social Psychology Quarterly, 48, 71-78.

Felson, R. B. (1989). Parents and the reflected appraisal process: A lon-gitudinal analysis. Journal of Personality and Social Psychology, 56,965-971.

Felson, R. B., & Reed, M. (1986). The effect of parents on the self-appraisals of children. Social Psychology Quarterly, 49, 302-308.

Festinger, L. (1957). A theory of cognitive dissonance. Evanston, IL:Row, Peterson.

Gecas, V. (1982). The self-concept. Annual Review of Sociology, 8, 1-33.

Goffman, E. (1959). The presentation of self in everyday life. GardenCity, NY: Doubleday-Anchor.

Gouldner, A. W. (1960). The norm of reciprocity: A preliminary state-ment. American Sociological Review, 25, 161-178.

Hixon, J. G., & Swann, W. B., Jr. (1993). When does introspection bearfruit? Self-reflection, self-insight, and interpersonal choices. Journalof Personality and Social Psychology, 64, 35-43.

Hoelter, J. W. (1984). Relative effects of significant others on self-evalu-ation. Social Psychology Quarterly, 47, 255-262.

Jones, S. C. (1973). Self and interpersonal evaluations: Esteem theoriesversus consistency theories. Psychological Bulletin, 79, 185-199.

Judd, C. M , & Kenny, D. A. (1981a). Estimating the effects of socialinterventions. Cambridge, England: Cambridge University Press.

Judd, C. M., & Kenny, D. A. (198 lb). Process analysis: Estimating me-diation in treatment evaluations. Evaluation Review, 5, 602-619.

Jussim, L. (1991). Social perception and social reality: A reflection-construction model. Psychological Review, 98, 54-73.

Jussim, L., Soffin, S., Brown, R., Ley, T, & Kohlhepp, K. (1992). Un-derstanding reactions to feedback by integrating ideas from symbolicinteractionism and cognitive evaluation theory. Journal of Personalityand Social Psychology, 62, 402-421.

Kenny, D. A., & DePaulo, B. M. (1993). Do people know how othersview them?: An empirical and theoretical account. Psychological Bul-letin, 114, 145-161.

Kinch, J. W. (1963). A formalized theory of the self-concept. AmericanJournal of Sociology, 68, 481-486.

Lecky, P. (1945). Self-consistency: A theory of personality. New York:Island Press.

Levesque, M. J., & Kenny, D. A. (1994). Accuracy of behavioral predic-

IDENTITY NEGOTIATION AND SELF 1021

tions at zero acquaintance: A social relations analysis. Journal of Per-sonality and Social Psychology, 65, 1178-1187.

Lundgren, D. C , Jergens, V. H., & Gibson, J. L. (1982). Marital power,roles, and solidarity and husbands' and wives' appraisals of self andother. Sociological Inquiry, 52, 33-52.

Manis, M. (1955). Social interaction and the self-concept. Journal ofAbnormal and Social Psychology, 51, 362-370.

McCrae, R. R., & Costa, P. T, Jr. (1987). Validation of the five-factormodel of personality across instruments and observers. Journal ofPersonality and Social Psychology, 52, 81-90.

Mead, G. H. (1934). Mind, self, and society. Chicago: University of Chi-cago Press.

Merton, R. K. (1948). The self-fulfilling prophecy. Antioch Review, 8,193-210.

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

Regan, D. T. (1971). Effects of a favor and liking on compliance. Journalof Experimental Social Psychology, 7, 627-639.

Rosenberg, M. (1965). Society and the adolescent self-image. Princeton,NJ: Princeton University Press.

Rosenberg, M. (1973). Which significant others? ^mm'ca« BehavioralScientist, 16, 829-860.

Rosenthal, R. (1976). Experimenter effects in behavioral research. NewYork: Appleton-Century-Crofts.

Rosenthal, R., & Jacobson, L. (1968). Pygmalian in the classroom:Teacher expectation and the pupils' intellectual development. NewYork: Holt, Rinehart & Winston.

Rosenthal, R., & Rubin, D. B. (1978). Interpersonal expectancy effects:The first 345 studies. Behavioral and Brain Sciences, 3, 377-386.

Schafer, R. B., & Keith, P. M. (1985). A causal model approach to thesymbolic interactionist view of the self-concept. Journal of Personal-ity and Social Psychology, 48, 963-969.

Secord, P. F , & Backman, C. W. (1965). An interpersonal approach topersonality. In B. Maher (Ed.), Progress in experimental personalityresearch (Vol. 2, pp. 91-125). San Diego, CA: Academic Press.

Shrauger, J. S., & Lund, A. (1975). Self-evaluation and reactions to eval-uations from others. Journal of Personality, 43, 94-108.

Shrauger, J. S., & Schoeneman, T. J. (1979). Symbolic interactionistview of self-concept: Through the looking glass darkly. PsychologicalBulletin, 86, 549-573.

Snyder, M., & Swann, W. B., Jr. (1978). Behavioral confirmation in so-cial interaction: From social perception to social reality. Journal ofExperimental Social Psychology, 14, 148-162.

Snyder, M., Tanke, E. D., & Berscheid, E. (1977). Social perception and

interpersonal behavior: On the self-fulfilling nature of social stereo-types. Journal of Personality and Social Psychology, 35, 656-666.

Sroufe, L. A., & Fleeson, J. (1986). Attachment and the construction ofrelationships. In W. Hartup & Z. Rubin (Eds.), Relationships anddevelopment (pp. 51-71). Hillsdale, NJ: Erlbaum.

Steele, C. M., Spencer, S. J., & Lynch, M. (1993). Self-image resilienceand dissonance: The role of affirmational resources. Journal of Per-sonality and Social Psychology, 64, 885-896.

Stryker, S. (1987). The vitalization of symbolic interactionism. SocialPsychology Quarterly, 50, 83-94.

Stryker, S., & Statham, A. (1985). Symbolic interaction and role theory.In G. Lindzay & E. Aronson (Eds.), Handbook of social psychology(Vol. 2, pp. 311-378). New York: Random House.

Swann, W. B., Jr. (1983). Self-verification: Bringing social reality intoharmony with the self. In J. Suls & A. G. Greenwald (Eds.), Psycho-logical perspectives on the self(Vo\. 2, pp. 33-66). Hillsdale, NJ:Erlbaum.

Swann, W. B., Jr. (1987). Identity negotiation: Where two roads meet.Journal of Personality and Social Psychology, 53, 1038-1051.

Swann, W. B., Jr. (1990). To be adored or to be known: The interplay ofself-enhancement and self-verification. In R. M. Sorrentino & E. T.Higgins (Eds.), Handbook of motivation and cognition: Foundationsof social behavior (Vol. 2, pp. 408-448). New York: Guilford Press.

Swann, W. B., Jr., & Ely, R. J. (1984). A battle of wills: Self-verificationversus behavioral confirmation. Journal of Personality and SocialPsychology, 46, 1287-1302.

Swann, W. B., Jr., Griffin, J. J., Predmore, S., & Gaines, B. (1987). Thecognitive-affective crossfire: When self-consistency confronts self-en-hancement. Journal of Personality and Social Psychology, 52, 881-889.

Swann, W. B., Jr., Hixon, J. G., & De La Ronde, C. (1992). Embracingthe bitter "truth": Negative self-concepts and marital commitment.Psychological Science, 3, 118-121.

Swann, W. B., Jr., Hixon, J. G., Stein-Seroussi, A., & Gilbert, D. T.(1990). The fleeting gleam of praise: Behavioral reactions to self-rele-vant feedback. Journal of Personality and Social Psychology, 59, 17-26.

Swann, W. B., Jr., & Shroeder, D. G. (1994). The search for beauty andtruth: A framework for understanding reactions to evaluative feedback.Manuscript submitted for publication.

Swann, W. B., Jr., Stein-Seroussi, A., & Giesler, B. (1992). Why peopleself-verify. Journal of Personality and Social Psychology, 62, 392-401.

Swann, W. B., Jr., Stein-Seroussi, A., & McNulty, S. E. (1992). Outcastsin a white lie society: The enigmatic worlds of people with negativeself-conceptions. Journal of Personality and Social Psychology, 62,618-624.

1022 SHAWN E. McNULTY AND WILLIAM B. SWANN, JR.

Appendix

Table A1Means, Standard Deviations, Ns, and Zero-Order Correlations (With Pairwise Deletion of Missing Data)by Trait for the Variables Measured in Study 1

Zero-order correlations

Variable M SD N

Academic abilityTl Self-viewTl AppraisalT2 Self-viewT2 AppraisalActual trait (SAT)

Social skillsTl Self-viewTl AppraisalT2 Self-viewT2 AppraisalActual traita

Athletic abilityTl Self-viewTl AppraisalT2 Self-viewT2 AppraisalActual trait (high school activities)

AttractivenessTl Self-viewTl AppraisalT2 Self-viewT2 AppraisalActual trait*

Neuroticismb

Tl Self-viewTl AppraisalT2 Self-viewT2 Appraisal

ExtraversionTl Self-viewTl AppraisalT2 Self-viewT2 Appraisal

OpennessTl Self-viewTl AppraisalT2 Self-viewT2 Appraisal

AgreeablenessTl Self-viewTl AppraisalT2 Self-viewT2 Appraisal

ConscientiousnessTl Self-viewTl AppraisalT2 Self-viewT2 Appraisal

Global worthTl Self-viewTl AppraisalT2 Self-viewT2 Appraisal

6.136.235.975.93

1,026.06

6.035.645.945.265.07

4.674.974.714.740.00

5.715.625.775.525.44

6.716.036.725.72

5.885.545.915.38

6.225.486.325.28

7.066.876.936.25

6.686.286.425.78

31.4535.7431.5034.42

1.171.141.201.17

152.31

1.401.451.451.441.05

1.811.781.871.500.71

1.361.351.291.171.28

1.741.741.711.43

1.701.781.791.61

1.511.761.501.88

1.561.291.611.64

1.621.431.581.63

7.054.167.454.58

6969696966

6969696958

6969696968

6969696958

69696969

69696968

69696969

69696969

69696969

69696867

1.00

1.00

1.00

1.00

1.00

1.00

1.00

1.00

1.00

1.00

.131.00———

.301.00—

—

.541.00—

—

.201.00—

—

.221.00

—

.241.00

—

.091.00

—

.021.00

—

.191.00

—

.061.00

.77

.101.00——

.66

.341.00

—

.83

.571.00

—

.77

.271.00

—

.70

.091.00—

.79

.321.00—

.71

.251.00—

.67

.141.00—

.41

.301.00—

.76

.151.00

.44

.54

.431.00—

.44

.63

.491.00—

.55

.62

.631.00—

.01

.67

.091.00—

.09

.33

.231.00

.39

.65

.511.00

.16

.80

.281.00

-.08.42.101.00

.11

.51

.321.00

-.02.52.041.00

.35

.20

.34

.361.00

-.16-.07-.05-.221.00

.49

.37

.45

.381.00

-.01-.03.17

-.111.00

Note. T1 = Time 1; T2 = Time 2; SAT = Standard Achievement Test.* Judged by raters. b This item was presented as emotional stability, so that higher numbers reflect less neuroticism.

IDENTITY NEGOTIATION AND SELF

Table A2Means, Standard Deviations, Ns, and Zero-Order Correlations (With Pairwise Deletion ofMissing Data) by Trait for the Variables Measured in Study 2

1023

Variable

Academic abilityTl Self-viewTl AppraisalT2 Self-viewT2 AppraisalT2 Reflected appraisalT2 Knowledgeability

Social skillsTl Self-viewTl AppraisalT2 Self-viewT2 AppraisalT2 Reflected appraisalT2 Knowledgeability

Athletic abilityTl Self-viewTl AppraisalT2 Self-viewT2 AppraisalT2 Reflected appraisalT2 Knowledgeability

AttractivenessTl Self-viewTl AppraisalT2 Self-viewT2 AppraisalT2 Reflected appraisalT2 Knowledgeability

Global worthTl Self-viewTl AppraisalT2 Self-viewT2 AppraisalT2 Reflected appraisalT2 Knowledgeability

M

6.056.065.875.975.014.88

6.165.846.185.744.924.92

4.744.424.914.674.014.63

5.635.625.655.524.515.08

32.3134.3733.0132.832.991.53

SD

1.171.141.291.141.081.51

1.411.471.351.541.401.63

1.761.721.681.751.671.74

1.321.261.181.261.261.40

6.085.436.386.321.121.19

N

959595959595

959594959595

959595959595

959595959595

959493949494

1

1.00

—

1.00

—

1.00

1.00

1.00

—

2

.211.00

.251.00

.481.00

.201.00

.051.00

—

Zero-order correlations

3

.63

.301.00

.63

.111.00

.82

.371.00

.70

.111.00

.83-.121.00

—

4

.30

.73

.381.00

.28

.71

.121.00

.55

.81

.441.00

.10

.72

.091.00

.07

.66

.011.00

—

5

.35

.23

.31

.321.00

.45

.33

.39

.461.00

.50

.41

.49

.501.00

.49

.22

.55

.151.00

.31

.06

.25

.071.00

6

.17

.33

.05

.35

.551.00

.31

.39

.22

.50

.601.00

-.01.03

-.00.13.17

1.00

.32

.28

.24

.20

.481.00

-.18-.10-.21-.23-.291.00

Note. T1 = Time 1; T2 = Time 2.

Received January 12, 1993Revision received April 18, 1994

Accepted April 22, 1994