: INTERPERSONAL RELATIONS AND GROUP PROCESSES Negotiating Social Identity When Contexts Change: Maintaining Identification and Responding to Threat Kathleen A. Ethier and Kay Deaux The impact of change in context on identity maintenance, the implications of maintenance efforts for group identification, and the effects of perceived threats to identity on self-esteem associated with group membership are examined in a longitudinal study of Hispanic students during their 1st year at predominately Anglo universities. Whereas ethnic identity is initially linked to the strength of the students' cultural background, maintenance of ethnic identity is accomplished by weakening that link and remooring the identity to the current college context. Results suggest 2 distinct paths by which students negotiate their ethnic identity in a new context. Students with initially strong ethnic identity become involved in cultural activities, increasing the strength of their identification. In contrast, students with initially weaker identification perceive more threat in the environment, show decreases in self-esteem associated with group membership, lowering identification with their ethnic group. The findings both support social identity theory and illustrate the need for more contextual analyses of identity processes. Identification of oneself with other people who share common attributes is an important aspect of self-definition. Theories of social identity, developed by Tajfel (1981), Turner (1987), and others, emphasize the importance of collective membership and the significant effects that group membership can have on behavior. These behaviors include feelings of attraction toward members of the in-group, stereotypic judgments of out-group members, social influence, and preferential treatment toward the in-group (Abrams & Hogg, 1990; Hogg & Abrams, 1988; Turner, 1987). As a number of investigators have shown, the need for and expression of social identity is not static. Brewer (1991), for example, found that the motivation to claim group membership depends on the competing needs for inclusiveness and uniqueness, whereby people seek an optimal level of distinctiveness in their choice of a collective. What particular identity is claimed can depend on situational cues that make an identity salient or that fit with one's own priorities (Deaux & Major, 1987; Oakes, 1987). Portions of these data were presented at the 98th Annual Convention of the American Psychological Association. Boston, 1990. Preparation of this article was facilitated by a grant from the National Science Foundation (BNS-9110130) to Kay Deaux. We thank Michelle Fine, Tracey Revenson, and members of the Identity Research Seminar at the City University of New York Graduate Center for their advice and commcntsonadraft of this article. Correspondence concerning this article should be addressed to either Kathleen A. Ethier, Department of Psychology, Yale University, P.O. Box 208205. New Haven, Connecticut 06520-8205 or Kay Deaux, City University of New York Graduate Center, 33 West 42 Street, New York, New York 10036-8099. Although the presentation of self may be quite variable, it can also be argued that the self-concept is generally stable across time. Individuals not only view themselves and others in consistent terms, but they actively create social situations that support their views of themselves (Swann, 1983). Across the life span, however, there are transitions that can have significant effects on self-definition (Hormuth, 1990; Ruble, 1994). During these transitions, individuals may find it necessary to adapt in some way to changes in environmental opportunities and demands. These adaptations may involve more than momentary responses to situational pressures; rather, the new situation may elicit fundamental changes in the meaning, importance, or support that a central identity has (Deaux, 1993). These issues of redefinition and change arc particularly interesting when one considers social identity categories that are ascribed to the person, such as ethnicity or gender, and might be thought to be particularly resistant to change. The importance of ethnic identity is well documented. Within the United States, the majority of people who are demographically designated as ethnic group members subjectively claim this identity as well (Waters, 1990). Ethnic identity is embedded in a multidimensional context, related to factors such as language, cultural background, geographic region, social class, and political conflict (Christian, Gadfield, Giles, & Taylor, 1976; Giles, Llado, McKiman, & Taylor, 1979; Giles, Taylor, & Bourhis, 1977; Giles, Taylor, Lambert, & Albert, 1976; Phinney, 1990). The importance of these links is suggested by Ethier and Deaux's (1990) finding that the strength of ethnic identity is significantly predicted by such factors as the language spoken in the home, the ethnic composition of the neighborhood, and the percentage of a student's friends who were in the same ethnic group. If a person's group identification is supported by a particular Journal of Personality and Social Psychology, 1994, Vol. 67. No. 2,243-23 I Copyright 1994 by ihe American Psychological Association. Inc. 00220514/94/S3.0O 243 244 KATHLEEN A. ETHIER AND KAY DEAUX context (Abrams, 1992; Ethier & Deaux, 1990), what happens when the person leaves that context and moves into another? According to social identity and self-categorization theories, contextual change that increases the salience of a particular identity leads to an increase in group identification (Emler & Hopkins, 1990; Oakes, 1987; Waddell & Cairns, 1986). Thus, in a study of Welsh identity, Christian et al. (1976) found that when group conflict was made salient by having subjects write essays about Welsh-English conflict, group identification was stronger. Studies of the effect of contextual change on social identity have generally been focused on short-term situational changes (Christian et al., 1976; Haslam, Turner, Oakes, & McGarty, 1992; Waddell & Cairns, 1986). Are the issues the same for long-term contextual change? By change in context, we do not mean a temporary shift in situation or place (e.g., having to perform a task that has some relation to a particular social identity, or a temporary move from one environment to another). Instead, our question concerns the effects on social identity of a complete change of environment, where the former supports of an identity (e.g., contact with group members) no longer exist as the person has known them. Because we are interested in looking at long-term change, the issues may be more complex, and the predictions of social identity theory and self-categorization theory may need to be expanded. Salience and Identity Change Social identity theory and self-categorization theory posit that when identity is made salient, as for example by a change in context, a person will become increasingly identified with his or her group. The concept of salience can be elusive, however, particularly when dealing with long-term changes in context. There are at least three bases on which one might make predictions about the influence of salience on group identification. First, one might posit that chronic levels of group identification would predict identity salience. Thus, those individuals who are more highly identified with their group would be more likely to experience that identity as salient, independent of situational context. In the specific context of Hispanic students entering college, we would then predict that students with a strong cultural background would be more likely to engage in activities relevant to their ethnic identification. A second saliency prediction is based on the contrast between a student's self-definition and the current context. Thus, as the research of McGuire and his colleagues shows, people whose status (e.g., ethnicity, gender, or hair color) is a minority in their group are more likely to be aware of that characteristic than those whose status is a majority (McGuire, McGuire, Child, & Fujioka, 1978). This position would suggest that ethnicity would be equally salient for all Hispanic students entering primarily Anglo universities. A third model of salience would consider the contrast between a student's past background and current context. If these two contexts were markedly different, salience should be greater than if there were no change. By this line of reasoning, students from primarily Hispanic neighborhoods and schools should find ethnicity more salient in the new primarily Anglo environ- ment than students whose previous background more closely resembled their current context. Thus, although each of these positions is consistent with the prediction from social identity theory that salience increases group identification, each proposes a different basis for determining when identity will be salient. When the change involves large-scale environments studied over an extended period of time, it must also be recognized that considerable variation exists. One specific situation within the general context (e.g., attending a relevant group meeting) may make an identity salient, whereas others (e.g., going to a history class) may not. These variations are in part dependent on individual choices and in part are a function of the opportunity structure available. Because of the possible variation in the natural environment, we would predict that individual differences in salience and identification should be particularly noticeable. As Abrams (1990) has argued, people often make choices about which course of action to follow, and social identity theory needs to take individual variation into account when considering the identity-salience relationship. Group involvement, both before the transition and in the new context, should be an important factor in determining whether an individual experiences increased salience in the new environment. Identity Maintenance and Change Social identity, as manifested in natural group memberships, cannot be conceptualized only as a cognitive categorization process. Rather, social identities are supported and sustained by a network of social relationships (Abrams, 1992). During times of transition, the relationship between the individual and his or her environment changes, and the person must adapt to those changes in some way (Hormuth, 1990). If a social identity is supported by a network of relationships, then one would anticipate that a change in context, for example, a change in physical location or a change in the social environment, would have some impact on identity. The ways in which the person had previously maintained the identity are no longer valid or useful in the new context, and the person must change the way in which he or she maintains the identity. Thus, in a new context, maintaining a social identity must include a process of remooring the identity to new social supports. Specifically, we hypothesize that to successfully maintain an identity in a new environment, a person must develop new bases for supporting that identity and, in the process, detach the identity from its supports in the former environment. Because any activity expended in developing these new links implies commitment to the identity, one could also predict that the process of remooring would in turn strengthen identification with the group (Aronson & Mills, 1959). Thus, although remooring is assumed to imply a decrease in the links to previous bases of identification, it should not result in any decrease in the strength of identification itself. Responses to Identity Threat New environments may challenge the meaning or value of an identity. Breakwell (1986) describes a variety of situations, including the loss of employment, the loss of a spouse, and cul- NEGOTIATING SOCIAL IDENTITY 245 tural conflicts, that pose threats to identity. If severe enough, threats can call into question the very existence of an identity. In other cases, the existence of the identity may not be threatened, but rather the meanings or value associated with the identity are questioned. The latter process is particularly likely when an identity is associated with membership in an ascribed category, for example, race, ethnicity, or gender. When experiencing threat to these identities, individuals are unlikely to consider abandoning the identity but may well alter the way that they feel about that identity. In other words, one's identification with the group does not change, but the self-esteem associated with that group becomes more negative. Consistent with this line of reasoning, Frable, Wortman, Joseph, Kirscht, & Kessler (1994) found that gay men who perceive more stigmatization of their group have lower self-esteem. There has been disagreement within social identity theory concerning the consequences of lowered self-esteem. According to social identity theory, the "need for positive self-esteem" (Turner, 1982, p. 33) is a fundamental human motivation. When an identity is made salient, this need is satisfied by positive evaluation of one's own group. However, what if a positive evaluation is not possible? Tajfel (1981) would predict that, if possible, the individual would leave the group; however, a person cannot easily leave an ascribed category. Hogg and Abrams (1990) pointed out the difficulty in finding support for the self-esteem hypothesis and suggested that other self-related motives (e.g., self-consistency) might override the motive for positive self-esteem. They argued that individuals are motivated to categorize themselves in the most meaningful way as dictated by the context. This process may result in a number of behaviors, including intergroup discrimination, acquiescence, intragroup normative competition, elevated self-esteem, or depressed self-esteem, and these behaviors should be dictated by the "socio-cultural and contextual factors" (p. 47). Thus, if it is not realistic to derive positive self-esteem from group membership, then other motives, such as the need to maintain a coherent and stable self-concept, could keep the individual from discarding the group membership. In other words, it might be more beneficial to live with depressed self-esteem than to lower group identification, especially if that group identification is difficult to forsake. On the other hand, if group identification is not strong, then the need for stability of the identity should also be weaker. In this case, the individual would be more likely to lower group identification to avoid further experiences of low self-esteem. In other words, the social identity theory prediction that lowered self-esteem will cause an individual to leave the group would be supported in a psychological sense, even if actual group membership could not be changed. Thus, we suggest that the relationship between self-esteem and group identification will vary depending on the importance of group membership. In minimal group situations, like those often used in social identity research, there is probably less variation in group identification than is found in natural groups. Although individuals may respond as members of the minimal group, these reactions do not mean that their identification with the group is strong or that group membership is central to the person. Thus, in these situations it would not be detrimental to the need for stability for the person to leave the group as a result of low self-esteem. By examining natural groups such as ethnic or racial groups in which individuals have spent their lives as group members, lived in communities with other group members, and spoken the same language as other group members, we can examine a situation in which different goals may have priority. In sum, we suggest that prior group involvement is an important factor in determining whether an individual experiences chronic salience, whether he or she becomes involved in the group in the new context, and whether he or she experiences the new context as threatening to his or her identity. Research Questions Ethier and Deaux (1990) explored both identity supports and perceptions of threat to ethnic identity in a study of Hispanic students who had begun their first year of university at primarily Anglo colleges. The present study extends those initial findings with a longitudinal investigation over the course of the students' first full year of college, investigating the relationships among threat, maintenance efforts, and change in a specific social identity. By interviewing the students at three times during their first year at college, we could examine the consequences for social identity in a naturally occurring situation that involves major contextual change. In so doing, we extend the range of investigation for social identity theories. We expected these students to adapt to their new environment in several ways. First, we expected to see general trends of remooring, that is that students would change the way they supported their Hispanic identity, moving from family and home culture to group involvement at school. Second, we expected that the students would make efforts to maintain stability, so that those students who were more involved in their ethnic group would be more likely to join Hispanic groups and have Hispanic friends at the university. We also expected that group involvement would be positively related to changes in ethnic identity. Third, we considered the possibility that students would react in different ways to the new environment as a function of the strength of their ethnic identification and prior background variables and that the relationship between threat, involvement, and self-esteem would be influenced by identification and cultural background. Method Sample and Recruitment Procedure Hispanic1 first-year students at two Ivy League universities were identified through lists provided either by the Dean of Students office at the university or through the university telephone directory. Sixty-five students were telephoned and asked to participate in a study about Hispanic students in their first year of college. All students who were contacted resided on the campus. Four students declined to participate; 16 agreed but did not appear for their first scheduled interview. Students who agreed to participate were scheduled for a first in- 1 Those students who did claim this identity used a variety of terms to label themselves including Latino, Chicana, Puerto Rican, and Mexican-American in addition to Hispanic. Without attempting to resolve the political issues involved here, we use Hispanic as the more inclusive term. 246 KATHLEEN A. ETHIER AND KAY DEAUX terview during late November or early December of their first semester at university. Two additional interviews were scheduled during the academic year. The second interview took place in February, shortly after students returned from the holiday break; the third was conducted in May as the students were completing their first year in college. All interviews took place in university buildings; each interview lasted approximately 45 min and was conducted in English. Students were paid $5 for their participation in each interview. Although the repeated testing necessitated keeping a record of names and questionnaires, these lists were confidential and kept separate from the data. A total of 45 students (28 men and 17 women) participated in the first interview. Of this total, 39 students (87%) participated at both Time 1 and Time 2 and 36 students (80%) were interviewed on all three occasions. The modal age of the students was 18 years, with a range of 17 to 19 years. All of the students were of Hispanic background, but there was a mixture of nationalities. Twenty-seven of the students classified themselves as Mexican-American. Most of these students were born in the United States; however, 6 were born in Mexico and moved to the United States before the age of 16. Seventeen students classified themselves as Puerto Rican. Of this number, 11 were born on the mainland and 6 were from Puerto Rico. One student was of Spanish descent. Measures The key concepts measured were identity, self-esteem associated with group membership, and perceptions of threat. Each of these variables was assessed at each of the three interviews. In addition, questions concerning past and present context for ethnic identity were included at each interview. Identity. Identity was assessed through a combination of qualitative and quantitative methods. First, to assess the pattern of identities by which individuals defined themselves, we asked each student to name all of the identities that were important to him or her. In the instructions for this task, they were given possible examples (e.g., age, gender, relationships to other people, and race or ethnicity). If any of a preestab-lished set of identities (Hispanic, student, son or daughter, family member, and friend) was not named voluntarily, the interviewer specifically asked the student whether he or she had these as part of the self-concept.2 Students were then asked to list the attributes or characteristics that they associated with each identity to understand the subjective meaning that students attached to their identities. The personal significance or importance of an identity was assessed in quantitative terms with two measures. First, students were asked to rate the importance of each identity that they named on a scale of 1 (not at all important) to 7 (very important). The second measure of group identification was the identity subscale of the Collective Self-Esteem Scale (Luhtanen & Crocker, 1992), with items written to refer to Hispanic identity specifically. This scale consists of four items, including "Being Hispanic is an important reflection of who I am," and "Being Hispanic is an important part of my self-image." As reported in Ethier and Deaux (1990), coefficient alpha for this scale at Time 1 was .92. Similar alpha coefficients were obtained for Times 2 and 3 (.91 and .91, respectively). Collective self-esteem. Self-esteem can be conceptualized in terms of personal self-esteem, referring to a general evaluation of one's individual self, or collective self-esteem*, referring to one's membership in collective groups. In this study, we define self-esteem in collective terms, referring to one's evaluation of one's ethnic group membership. Self-esteem was assessed using the Private Acceptance subscale of Luhtanen and Crocker's (1992) Collective Self-Esteem Scale. Again, items were phrased in terms of Hispanic identity, specifically. This subscale consists of four items, including "In general, I'm glad that I'm Hispanic" and "I often regret that I am Hispanic." As reported in Ethier and Deaux (1990), coefficient alpha for this scale at Time 1 was .66. Identical alpha coefficients were obtained for Times 2 and 3. Perceptions of threat. Perceptions of threat to one's ethnic identity were assessed by a six-item scale developed by us. This scale included items such as "I feel that my ethnicity is incompatible with the new people I am meeting and the new things I am learning" and "I can not talk to my friends at school about my family or my culture."4 Students answered each item on a 7-point scale, indicating whether they had experienced that feeling or situation (1 = not at all, 1 = a great deal). The possible range of scores was 6 to 42; the obtained range was 6 to 41. Coefficient alpha for Times 1, 2, and 3 were .66, .82, and .72, respectively. Ethnic involvement. At each interview, a variety of questions were asked to assess the extent of involvement with family, friends, and Hispanic culture. In the initial interview, these questions were directed at community and family background. Responses to six questions were combined to form an index termed Strength of Cultural Background (SCB; a = .80). Items included in this index referred to mother's and father's birthplace, language spoken in the home, percentage of home community that was Hispanic, and percentage of Hispanic high school friends. At the second and third interviews, questions were directed at the level of students' involvement in ethnic culture at school. At Time 2, these questions referred to involvement during the first semester, and at Time 3, they referred to involvement during the second semester. Again, composite indexes were developed. Hispanic activity consisted of two items, rs(38) = .70 at Time 2 and .74 at Time 3, that asked about participation in Hispanic activities at college and percentage of current friends who were Hispanic. Results Stability of Measures Table 1 presents the means and standard deviations for each of the major variables in the study at each of the three occasions. Table 2 shows the correlations of each measure across the three occasions of testing. In the case of the identities, common element correlations (McNemar, 1962) were used to determine the repetition of elements across time for qualitative data. High correlations in this case indicate that students were naming a similar set of identities on the different occasions. As Table 1 shows, there were no mean changes in the number of identities that students mentioned, the importance that they attached to their Hispanic identity, or the level of self-esteem associated with ethnic identity. Perceptions of threat to the Hispanic identity did change during the course of the year, however, .F(2, 70) = 2.91, p < .06. A trend analysis shows a significant curvilinear trend in the data, F\l, 102) = 31.25,/>< .01. As the means in Table 1 show, perceived threats were the strongest at 2 Although this probe increases the frequency with which a give identity is mentioned, it by no means results in universal endorsemen In our sample, 13% of the students did not claim this identity, even affc being prompted. 3 We use the term collective self-esteem to refer to self-esteem assoc ated with group membership. Luhtanen and Crocker's Collective Sel Esteem Scale actually measures four different dimensions of feelinj associated with group membership (membership, private acceptanc public acceptance, and identification). In the present study, we use tl Private Acceptance subscale, which refers to personal feelings about tl group, to index collective self-esteem. 4 See Ethier and Deaux (1990, p. 433) for a full listing of the items ( the Perceived Threat scale. NEGOTIATING SOCIAL IDENTITY 247 Table 1 Means and Standard Deviations of Major Variables Time Variable 1 2 3 Number of identities M 7.6 6.8 7.4 SD 2.4 1.8 1.6 Hispanic importance M 4.4 4.4 4.6 SD 2.5 2.5 2.3 Identification M 19.5 20.4 19.3 SD 6.4 5.6 5.5 Self-esteem M 24.1 23.9 23.8 SD 3.6 3.2 3.3 Perceptions of threat M 12.1 13.1 12.7* SD 5.4 7.9 5.8 * p < .06. Time 2, soon after students had returned from holiday break and were beginning the second semester, but threat did not diminish substantially over the course of the second semester. As evident in Table 2, all of the measures were highly stable with the exception of self-esteem associated with group membership and, to a lesser extent, perceptions of threat. In both cases, the correlations are lowest for the longest interval of time, from early fall to late spring, as would be expected. In the case of collective self-esteem, the low correlation suggests considerable individual variation despite a constant mean level of collective self-esteem within the sample. Characteristics of Hispanic Identity The vast majority of the students mentioned Hispanic as one of their important identities (87%, 83%, and 86% at Times 1, 2, and 3, respectively). Although not surprising, given the explicit focus of the study, it does verify the central role that ethnicity plays for many people. Ethnicity was among the most important identities, although several others were similar in frequency of mention (e.g., student, friend, and daughter or son). In fact, both student and friend were mentioned as an identity by 100% Table 2 Stability of Variables Across Testing Occasions Testing occasions Variable 1-2 1-3 2-3 Identities named .69 .70 .79 Hispanic importance .75 .70 .96 Identification .72 .82 .74 Self-esteem .59 .42 .74 Perception of threat .78 .58 .70 Note. N = 39 for the Occasion 1-2 comparison and TV = 36 for the other comparisons. of the students at Times 2 and 3, whereas ethnicity was still not endorsed by some students. Furthermore, although the rated importance of Hispanic identity was stable and high, it was not as high as the rated importance of either the student identity (Ms = 5.3, 5.7, and 5.6, respectively) or the friend identity (Ms = 5.3, 5.9, and 6.0, respectively). As expected, the rated importance of Hispanic identity was significantly related to the identity subscale of the Collective Self-Esteem Scale, r(45) = .79, p < .01, at Time 1; i-(38) = .59, p < .01, at Time 2; and r(34) = .73, p<. 01, at Time 3. Students used many different terms to describe their Hispanic identity. As shown in Table 3, at all three times, positive feelings about the group (e.g., proud, aware, and loyal) and background characteristics (e.g., culture, language, family, and tradition) were mentioned most often. Background became significantly less prominent over time (Cochran's Q = 7.05, p < .03), dropping from 63% and 71%, at Times 1 and 2, to 48% by the end of the year. There is also a significant decrease (Cochran's Q = 14.0, p < .001) in negative feelings as well as in the use of personality attributes to characterize Hispanic identity (Cochran's Q = 18.3, p < .001). These qualitative changes support our view that although objective membership in the group remains stable, the subjective meaning associated with the identity is open to change. Maintaining Identity Through Remooring We predicted that in response to the changes in environment that the students experienced, they would change the ways in which they maintained their Hispanic identity, remooring the identity within the new context. These expectations were sup- Table 3 Categories of Attributes Most Often Associated With Hispanic Identity Percentage of students Category Time 1 Time 2 Time 3 Positive feelings about the group (e.g. proud, loyal) 57.9 51.6 51.6 Positive feelings because of membership (e.g., happy, lucky) 12.2 25.8 19.4 Others' negative reactions (e.g., prejudice) 36.8 32.3 25.8 Negative feelings (e.g., resentment, doubt) 44.7 12.9 25.8** Active reactions (e.g., educate others) 15.8 29.0 29.0 Background (e.g., language, family) 63.2 71.0 48.4* Personality characteristics (e.g., caring, strong) 47.4 12.9 16.1** Identification with the group (e.g., "who I am") 21.1 38.7 32.3 Change (e.g., learning, growing) 23.7 12.9 12.9 Change over time: *p < .05. **/><.001. 248 KATHLEEN A. ETHIER AND KAY DEAUX ported. At the beginning of the year, the importance of the student's ethnic identity was significantly related to family background, both in terms of degree of Hispanic influence, r(43) = .42, p < .01, and to the specific importance of their identity as a son or daughter, ^43) = .30, p < .05.5 In contrast, at the second and third interviews, the link between cultural background and the strength of ethnic identification was no longer significant (rs = .26 and .19, respectively), suggesting that ethnic identity had been detached from its previous cultural context. The link between the importance of the identity as a son or daughter and Hispanic identity also loses its significance, although this detachment takes longer. At Time 2, ethnic identification and the importance of one's identity as a son or daughter remains im->f portant, r(38) = .34, p < .05. By Time 3, however, this relation- ship is no longer significant, r{34) = . 14, p < .25. Involvement with Hispanic activities on the campus appears to take the place of family background as a support for ethnic identity. At Time 2, the correlation between strength of identification and involvement in Hispanic activities during the preceding semester is significant, r(38) = .55, p < .01. This relation shows increased strength at Time 3, when ethnic identity is correlated with Hispanic activities during the second semester, r(34) = .62,p<.01. li Ethnic Involvement and Perceived Threat Ethnic involvement played a powerful but shifting role across the students' first year at university. As a prior condition, ethnic involvement was measured by the Strength of Cultural Back-Jj ground index. Initially, this background served to buffer stu- dents from perceived threats to their ethnic identity. Thus, SCB was negatively related to Perceived Threat (r(43) = -.39, p < .01), such that the stronger the students' cultural background, the less likely they were to perceive the new environment as threatening. Past ethnic involvement, as assessed by SCB, also predicted a student's tendency to become involved in Hispanic activities on the campus (r(38) = .41, p < .01). The more involved the students were in their ethnic group before the transition to college, the more likely they were to become involved in ethnic activities on campus. The importance of these indexes of ethnic involvement to group identification is evidenced in the results of regression analyses, used to determine the influence of involvement on changes in the strength of identification. In all of the regression analyses presented here, the initial value of the dependent variable (e.g., identification at Time 2) was entered into the equation first to control for any variance in the change score that could be explained by the initial value of the variable itself. These analyses show, first, that ethnic involvement before college (as measured by SCB) was a significant predictor of changes in identification from Time 1 to Time 2 (b = -.28, p < .05): The stronger the students' ethnic background, the stronger the identification with the group became during the student's first semester at college. Second, ethnic involvement at the university (as measured by the Hispanic activities index) was a significant predictor of changes in identification from Time 2 to Time 3 (b = -.65, p < .001): The more involved the students were with their ethnic group at college, the stronger their ethnic identity became in the second semester of the year. In fact, after controlling for the initial value of identification, Hispanic activities, by itself, explained 30% of the variance in changes in identification from Time 2 to Time 3. When ethnic involvement was not strong, the processes of identity negotiation were quite different. As suggested above, students whose ethnic involvement before college was low perceived the college environment as more threatening. Perceptions of threat, in turn, had negative effects on self-esteem. Perceived threat at Time 1 was a significant predictor of changes in self-esteem both from Time 1 to Time 2 (b = .36, p < .01) and from Time 2 to Time 3 (b = .42, p < .05). (There was no association between ethnic involvement and self-esteem.) In support of the prediction made by social identity theory, we find that collective self-esteem at Time 2 predicts changes in identification from Time 2 to Time 3 (b = -.32, p < .05). Thus, if students experience low self-esteem associated with their group membership, the strength of that identification decreases. Ethnic involvement before college moderates this relationship between collective self-esteem and identification. The interaction between SCB and private acceptance significantly predicts change in identification from Time 2 to Time 3 (b = .15, p < .05), such that students who had lower cultural background before college and who had lower self-esteem associated with group membership showed more negative changes in identification. Two Paths of Identity Negotiation These results suggest that there are two quite different processes occurring as students negotiate their ethnic identity in a changed context. For students who come in with a strong ethnic background, choices are made that continue ethnic involvement and result in a strengthening of the group identification. In contrast, students with a weak ethnic background show more signs of stress with resultant lower self-esteem and negative changes in identification. To explore this possibility further, we divided the sample into two groups, based on the initial strength of their ethnic involvement, as assessed by the SCB. Because the distribution of scores on this measure was essentially bimodal, we divided the sample at the point of separation. High ethnic involvement subjects scored between 6 and 12 on the SCB measure; low involvement subjects scored between 16 and 22. As would be expected on the basis of the previous findings, these two groups differed significantly on their involvement in Hispanic activities at school (/ = -2.50, p < .01) and in the degree of threat they perceived in the college environment (/ = 2.54, p < .01). Students who were more ethnically involved before the transition were more likely to become involved in ethnic activities at school and were less likely to experience threats to their identity, as compared with those students who were less ethnically involved before the transition. More interesting in terms of supporting a position that there 5 As reported in Ethier and Deaux (1990), the pattern of these relationships differed for men and women. General Hispanic background was the more significant predictor for men, whereas the specific parent-daughter link was more influential for women. NEGOTIATING SOCIAL IDENTITY 249 are two distinct paths of identity negotiation is the fact that SCB group (high vs. low) moderates the relationship between perceived threat and changes in self-esteem. The interaction between SCB group and perceived threat is a significant predictor of changes in private acceptance from Time 2 to Time 3 (b = .20, p < .03). Thus, those students with lower ethnic involvement before college and higher perceived threat were more likely to show negative changes in self-esteem later in the year. Discussion This study provides several unique insights on the process by which people negotiate a social identity when the context for enacting that identity has substantially changed. By focusing on an identity of considerable importance, that is, ethnic identity, and by tracking the students over the course of more than 6 months, we gained a measure of ecological validity that studies of social identity rarely attain. One key aspect of maintaining an identity when confronted with a new environment is a process we term remooring. In a general sense, ethnic identity is quite stable, particularly when one considers the frequency with which the category is endorsed by the individual. (Stability is not invariant, however, as Waters [1990] has shown.) What allows this stability to be maintained, however, is the process of remooring the identity to supportive elements in the new environment. Initially, a strong ethnic identity was supported by family background, by high school friends, and by neighborhood context. With the movement to a new locale, students who wished to maintain a strong ethnic identity needed to develop a new base of support for that identity, much as Hormuth (1990) has suggested. Students in the present study did this by linking their identity to people and activities on the campus that were consistent with a Hispanic identity. It is striking that these new links did not act as a supplement to cultural background but actually replaced the earlier basis of support. We suspect that this process is particularly characteristic of identities, like ethnicity, that are not directly tied to specific role relationships. A second important finding concerns the way in which previous group involvement shapes the individual's approach to an altered context. In this study, the students' history of ethnic involvement predicted the degree to which they made efforts to maintain their group membership. Students who came from communities with high concentrations of Hispanics, who spoke Spanish in their homes, and who had a high percentage of Hispanic friends in high school were more likely to join Hispanic organizations at college and to make friends with other Hispanic students. The students who made these efforts showed an increase in Hispanic identification; those who did not make these efforts showed a decrease in Hispanic identification. Social identity and self-categorization theory posit that an increase in salience will lead to an increase in identification. As discussed earlier, a clear statement about salience is not easily made. One might assume that Hispanic identity would be salient for all students in this study, given their clear minority status within the college population. If so, however, not all students responded to this salience with increased group identification. Only those students for whom ethnic involvement was high initially showed the increase in identification that social identity theory would predict. The obtained pattern of results would also be consistent with an assumption of salience based on the contrast between past context and current context. From this perspective, it would again be the students from the most Hispanic backgrounds for whom ethnic identity would be most salient on the college campus. One hesitates to infer, however, that ethnic identity was not salient for those students whose ethnic background was weaker. For these students, however, the awareness of ethnicity appears to have had negative effects, rather than the increased identification predicted by social identity theory. These findings are consistent with Abrams's (1990) contention that there may be considerable variation in people's actions once an identity is made salient. Whereas social identity theory has traditionally given little attention to individual variation, preferring to stress the common response to conditions of salience, the present results stress the need to consider more agentic possibilities by people acting in their natural environments. We have shown clearly, and indeed the finding makes a great deal of intuitive sense, that variations in the level of previous group involvement determine the effect of contextual change on ethnic identity. A third key finding from this study involves the reaction of students to their perception that the new environment was threatening to their identity. Our results suggest that in the face of threats, evaluations associated with the identity are particularly vulnerable to negative change. Students who perceived substantial threat and ambivalence about their identity as Hispanics showed subsequent drops in self-esteem associated with that identity. Thus, perceptions of threat to ethnic identity assessed in November correlated with negative changes in collective self-esteem from early in the year to midyear and from midyear to the end of the year. We also find strong support for Tajfel's (1981) prediction that low self-esteem associated with a particular group membership will lead the individual to move away from the group. Self-esteem associated with Hispanic identity was significantly correlated with changes in identification: Those students who evaluated their group negatively lowered their identification with the group, whereas those students who felt positively about their group showed an increase in identification. The observation that students take one of two paths in the new environment on the basis of previous group involvement is a particularly intriguing one. As discussed earlier, these students varied in the degree to which they were involved in their ethnic group before the transition. Students with a strong cultural background were more likely to become involved in their ethnic group at college and, subsequently, to show increases in identification with that group. In contrast, students without a strong cultural background were more likely to perceive threats to their identity, to have lower self-esteem associated with that identity, and to lower their identification with the group. There could be a number of underlying reasons for this pattern. One possibility is that the students with a strong cultural background are simply continuing to manage their ethnicity in the same ways they had before. They may be responding to the upheaval of leaving home and going to college by seeking out others with similar backgrounds as a way of making themselves more comfortable. Indeed several of the students in open-ended interviews 250 KATHLEEN A. ETHIER AND KAY DEAUX conducted at the end of the year mentioned that having a group of people who spoke the same language and had similar experiences was a benefit of being Hispanic and made adjustment to college easier for them. These highly identified students might also be more likely to use the supportive services made available by the university for minority students, such as ethnic counselors, cultural houses, and special orientations. In taking advantage of these opportunities, the highly identified students may have strengthened their identification and minimized perceptions of threat and consequent loss of self-esteem associated with group membership. The contrasting case of students who did not come from a strong ethnic background suggests quite different events and processes. It seems quite likely that these students felt conflict about being categorized as Hispanic by the university when they themselves did not strongly identify with their ethnic group. Because they are less likely to speak the language or to come from areas densely populated by other Hispanics, they may not feel accepted by other ethnic group members, thus failing to gain the support that the group might offer. Similarly, these students might be less likely to use the ethnically oriented services that the university provides. The end result is a decrease in ethnic group identification but a drop in self-esteem associated with group membership as well. Whatever the underlying dynamics, it is quite significant that students who are more strongly identified with their ethnic group fare better during this transition than the students who are less strongly identified with the group. This finding speaks to the protective nature of group identity in situations in which the group is a numerical minority and is possibly faced with discrimination from the majority group. Although our sample is small, the patterns are sufficiently strong to warrant confidence in the results. At the same time, we do not know how specific these phenomena are to elite private institutions in which a particular ethnic group, such as Hispanics, is such a numerical minority. When one's group is in the majority or even a substantial minority, environmental demands and opportunities could be quite different. Ethnicity might be less salient in such situations, creating fewer threats for those with weak ethnic identification and less perceived need to get involved in specific ethnic activities for those with high ethnic identification. Beyond the specific results, this study attests to the importance of studying identity-related experience in a natural field setting over the course of time (Deaux, 1993). This approach allows us to examine the dynamics of identity work—the rein-terpretation, reevaluation, and reconnection that allow identities to maintain seeming consistency over time. It also shows how theories of social identity play out with natural groups in realistic environments and how existing theories about social identity can be enriched and extended. References Abrams, D. (1990). How do group members regulate their behavior? An integration of social identity and self-awareness theories. In D. Abrams & M. A. Hogg (Eds.), Social identity theory: Constructive and critical advances (pp. 89-112). New York: Springer-Verlag. Abrams, D. (1992). Processes of social identification. In G. M. Break-well (Ed.), Social psychology of identity and the self-concept (pp. 57-99). London: Surrey University Press. Abrams, D., & Hogg, M. A. (1990). Social identity theory: Constructive and critical advances. New York: Springer-Verlag. Aronson, E., & Mills, J. (1959). The effect of severity of initiation on liking for a group. Journal of Personality and Social Psychology, 59, 177-181. Breakwell, G. (1986). Coping with threatened identities. London: Me-thuen. Brewer, M. B. (1991). The social self: On being the same and different at the same time. Personality and Social Psychology Bulletin, 17, 475-482. Christian, J., Gadfield, N. J., Giles, H., & Taylor, D. M. (1976). The multidimensional and dynamic nature of ethnic identity. International Journal of Psychology, 11, 281-291. Deaux, K. (1993). Reconstructing social identity. Personality and Social Psychology Bulletin, 19, 4-12. Deaux, K. (1993). Enacting social identity: Maintaining stability and dealing with change. In S. Stryker (Eds.), Self and affect in society. Deaux, K, & Major, B. (1987). Putting gender into context: An interactive model of gender-related behavior. Psychological Review, 94, 369-389. Emier, N., & Hopkins, N. (1990). Reputation, social identity and the self. In D. Abrams & M. A. Hogg (Eds.), Social identity theory: Constructive and critical advances (pp. 113-130). New York: Springer-Verlag. Ethier, K. A., & Deaux, K. (1990). Hispanics in ivy: Assessing identity and perceived threat. Sex Roles, 22, 427-440. Frable, D. E. S., Wortman, C, Joseph, J., Kirscht, J., & Kessler, R. (1994). Predicting self-esteem, well-being, and distress in a cohort of gay men: The importance of cultural stigma and personal visibility Manuscript submitted for publication. Giles, H, Llado, N., McKirnan, D. J., & Taylor, D. M. (1979). Social identity in Puerto Rico. International Journal of Psychology, 14, 185- 201. Giles, H, Taylor, D. M., & Bourhis, R. Y. (1977). Dimensions of Welsh identity. European Journal of Social Psychology, 7, 165-174. Giles, H, Taylor, D. M., Lambert, W. E., & Albert, G. (1976). Dimensions of ethnic identity: An example from northern Maine. Journal of Social Psychology, 100, 11-19. Haslam, S. A., Turner, J. C, Oakes, P. J., & McGarty, C. (1992). Context-dependent variation in social stereotyping: I. The effects of intergroup relations as mediated by social change and frame of reference. European Journal of Social Psychology, 22, 3-20. Hogg, M. A., & Abrams, D. (1988). Social identifications. London: Routledge. Hogg, M. A., & Abrams, D. (1990). Social motivation, self-esteem and social identity. In D. Abrams & M. A. Hogg (Eds.), Social identity theory: Constructive and critical advances (pp. 28-47). New York: Springer-Verlag. Hormuth, S. E. (1990). The ecology of the self: Relocation and self-concept change. Cambridge, England: Cambridge University Press. Luhtanen, R., & Crocker, J. (1992). A collective self-esteem scale: Self-evaluation of one's social identity. Personality and Social Psychology Bulletin, 18, 302-318. McGuire, W. J., McGmre, C. V, Child, P., & Fujioka, T. (1978). Salience of ethnicity in the spontaneous self-concept as a function of one's ethnic distinctiveness in the social environment. Journal of Personality and Social Psychology, 36, 511 -520. McNemar, Q. (1962). Psychological statistics. New York: Wiley. Oakes, P. J. (1987). The salience of social categories. In J. C. Turner (Ed.), Rediscovering the social group: A self-categorization theory(pp. 117-141). Oxford, England: Basil Blackwell. Phinney, J. S. (1990). Ethnic identity in adolescents and adults: Review of research. Psychological Bulletin, 108, 499-514. Ruble, D. (1994). A phase model of transitions: Cognitive and motiva- NEGOTIATING SOCIAL IDENTITY 251 tional consequences. In M. Zanna (Ed.), Advances in social psychology (PP- 163-214). San Diego, CA: Academic Press. Swann, W. B. (1983). Self-verification: Bringing social reality into harmony with the self. In J. Suis&A.G. Greenwald (Eds.), Psychological perspectives on the self (pp. 33-66). Hillsdale, NJ: Erlbaum. Tajfel, H. (1981). Human groups and social categories. Cambridge, England: Cambridge University Press. Turner, J. C. (1982). Towards a cognitive redefinition of the social group. In H. Tajfel (Ed.), Social identity and intergroup relations (pp. 15-40). Cambridge, England: Cambridge University Press. Turner, J. C. (1987). Rediscovering the social group: A self-categorization theory. Oxford, England: Basil Blackwell. Waddell, N., & Cairns, E. (1986). Situational perspectives on social identity in Northern Ireland. British Journal of Social Psychology, 25, 25-31. Waters, M. (1990). Ethnic options: Choosing identities in America. Berkeley: University of California Press. Received November 5, 1992 Revision received December 13, 1993 Accepted January 3,1994 ■ P&C Board Appoints Editor for New Journal: Journal of Experimental Psychology: Applied In 1995, APA will begin publishing a new journal, the Journal of Experimental Psychology: Applied. Raymond S. Nickerson, PhD, has been appointed as editor. Starting immediately, manuscripts should be submitted to Raymond S. Nickerson, PhD Editor, JEP: Applied Department of Psychology Tufts University Medford, MA 02155 The Journal of Experimental Psychology: Applied will publish original empirical investigations in experimental psychology that bridge practically oriented problems and psychological theory. The journal also will publish research aimed at developing and testing of models of cognitive processing or behavior in applied situations, including laboratory and field settings. Review articles will be considered for publication if they contribute significantly to important topics within applied experimental psychology. Areas of interest include applications of perception, attention, decision making, reasoning, information processing, learning, and performance. Settings may be industrial (such as human-computer interface design), academic (such as intelligent computer-aided instruction), or consumer oriented (such as applications of text comprehension theory to the development or evaluation of product instructions).