Developing the Ethnic Identity Scale Using Eriksonian and Social Identity Perspectives

Transcription

1 IDENTITY: AN INTERNATIONAL JOURNAL OF THEORY AND RESEARCH, 4(1), 9 38 Copyright 2004 Lawrence Erlbaum Associates, Inc. Developing the Ethnic Identity Scale Using Eriksonian and Social Identity Perspectives Adriana J. Umaña-Taylor, Ani Yazedjian, and Mayra Bámaca-Gómez Department of Human and Community Development University of Illinois at Urbana-Champaign Two studies were conducted to develop and explore the psychometric properties of the newly developed Ethnic Identity Scale (EIS). Consistent with Erikson s and Tajfel s theoretical perspectives, the EIS assesses 3 domains of ethnic identity formation: exploration, resolution, and affirmation. In both studies, participants (N = 846) completed measures of familial ethnic socialization and self-esteem in addition to completing the EIS. In Study 1, we employed exploratory and confirmatory analyses to examine, refine, and confirm the factor structure of the EIS among university students (n = 615). In Study 2, we examined the psychometric properties of the EIS among high school students (n = 231). Results revealed a three-factor solution that reflected the proposed components of exploration, resolution, and affirmation. Furthermore, the three subscales were related in expected ways to measures of familial ethnic socialization and self-esteem. Identity development involves the process of defining oneself as a group member within a broader social context and also serves as the framework that provides individuals with a coherent sense of self (Grotevant, 1992; Josselson, 1994). Researchers argue that identity formation is a critical developmental task faced during adolescence, the resolution of which serves as a guiding framework in adulthood (Josselson, 1994; Spencer, Swanson, & Cunningham, 1991; Swanson, Spencer, & Petersen, 1997). Although individuals social selves are composed of a number of component social identities (e.g., gender, occupation), in ethnicity-conscious societies like the U.S., it is imperative to more sufficiently ad- Requests for reprints should be sent to Adriana J. Umaña-Taylor, University of Illinois, Department of Human and Community Development, Urbana, IL, umana@uiuc.edu

2 10 UMAÑA-TAYLOR, YAZEDJIAN, BÁMACA-GÓMEZ dress issues of ethnic identity formation and its relation to the broader social self. This article examines the evolution of current conceptualizations of ethnic identity and introduces a measure that assesses its multifaceted nature. Importantly, individuals distinguish themselves from one another based on a wide range of categories. For some, ethnicity may be a useful categorization when dealing with the experiences of daily living. An ethnic group represents a group of people who maintain a subjective belief in their common descent and shared history, and who share certain cultural traits such as dress, art, music, food, literature, and language (Branch, 1999; Levine, 1997). Existing research suggests that ethnic identity is positively associated with important outcome variables such as individuals strategies for coping with discrimination (Chavira & Phinney, 1991; Phinney & Chavira, 1995) and their psychological well-being (see Umaña-Taylor, Diversi, & Fine, 2002, for a review). However, findings vary based on the conceptualization and, consequently, the measurement of ethnic identity (Phinney, 1995; Umaña-Taylor et al., 2002). As such, it is critical to develop valid and reliable methods with which to measure this construct. THEORETICAL BACKGROUND Much of the developmental work that has examined ethnic identity among adolescents has been grounded in both Tajfel s social identity theory and Erikson s theory of identity formation (e.g., Phinney, 1989; Umaña-Taylor & Fine, 2002). Here, we focus briefly on these two perspectives and then specifically discuss Marcia s operationalization of Erikson s theory and Phinney s application of these theories to ethnic identity development during adolescence. Social Identity Theory Social identity theory (Tajfel, 1981) posits that identity develops from both an individual s sense of belonging to a particular group and the affective component accompanying that sense of group membership. Furthermore, Tajfel suggested that individuals self-esteem is derived from their sense of group belonging and, consequently, those who maintain favorable definitions of group membership will also exhibit positive self-esteem (Phinney, 1992; Phinney, Cantu, & Kurtz, 1997). However, if the social climate in which individuals lives are embedded does not place value on the ethnic group, and individuals experience discrimination or prejudice, they may display lower self-esteem than members of groups who do not have these experiences. Ego Identity Formation Alternatively, Erikson s (1968) identity formation theory posits that identity development occurs through a process of exploration and commitment to important

3 ETHNIC IDENTITY SCALE 11 identity domains of a broader self-concept. However, Erikson s postulations do not articulate that one s commitment to a component identity is necessarily always positive. Rather, Erikson indicates that individuals will, as a result of exploration, resolve their feelings about the role of a particular component identity (e.g., vocational, religious, sexual, political) within their broader social self. Furthermore, Erikson s theory suggests that the culmination of such a period of exploration will lead the individual to reconcile his conception of himself and his community s recognition of him (Erikson, 1959, p. 120). In other words it is only through the process of exploration that individuals can come to a resolution regarding a particular identity. Thus, from an Eriksonian perspective, there are two critical components to the process of identity formation: exploration and commitment. Although social identity theory focuses more on the affective components of identity and how they are related to outcomes, Erikson s theory places greater emphasis on the process of identity development. Operationalization of Erikson s Theory Marcia s (1980, 1994) operationalization of Erikson s theory of identity formation allows researchers to classify individuals, based on their degree of exploration and commitment, into one of four identity statuses: diffuse, foreclosed, moratorium, and achieved. According to this typology, individuals who have not explored or committed to an identity would be considered diffuse and those who have explored but have not yet committed would be considered to be in moratorium. In contrast, individuals who have not explored, but have committed to a particular identity, would be considered foreclosed, whereas those who have both explored and committed would be considered achieved. This operationalization has been examined with regard to sex differences (Kroger, 1986), psychological well being (Josselson, 1994), and status differences on numerous psychological outcomes and personality characteristics (for reviews see Archer, 1989; Marcia, 1980; Serafini & Adams, 2002). Phinney s Three-Stage Model of Ethnic Identity Phinney s (1989) work on ethnic identity formation has drawn on Tajfel s (1981) social identity theory, Erikson s (1968) theory of global identity development, and Marcia s operationalization of Erikson s theory. In her early work, interviews were conducted with White, Asian, Hispanic, and Black adolescents to explore the application of Marcia s operationalization to the construct of ethnic identity. Because Phinney (1989) was unable to distinguish adolescents who were foreclosed with respect to their ethnicity from those who were diffuse, she developed a model that described ethnic identity development as a three-stage process, beginning with a lack of exploration of ethnicity (i.e., unexamined) and ending with resolution (i.e., achieved). During the unexamined stage, individuals may either have

4 12 UMAÑA-TAYLOR, YAZEDJIAN, BÁMACA-GÓMEZ adopted the dominant culture s characterization of their ethnic group or may have never been exposed to any groups other than their own. It is not until the second stage, moratorium, that individuals begin a period of exploration regarding their ethnicity. Such exploration may lead individuals to develop a meaning for their ethnicity in their own lives and define the salience of their ethnic identity with respect to their broader social self (Bachay, 1998; Phinney, 1990). During the final stage, achievement, individuals resolve their feelings about their ethnic identity and internalize the meaning of their ethnicity in their lives. It is at this point that individuals come to terms with group membership (Phinney, 1989, 1990). Consequently, Phinney s theoretical model conceptualizes ethnic identity achievement as a continuous variable, ranging from the lack of exploration and commitment to evidence of both exploration and commitment (Phinney, 1992, p. 161). Implications of Current Operationalizations of Ethnic Identity To assess her theoretical model of ethnic identity development, Phinney (1992) created the Multigroup Ethnic Identity Measure (MEIM), which examines individuals achievement (i.e., degree of exploration and commitment), ethnic behaviors (i.e., degree of participation in cultural activities), and affirmation and belonging (i.e., degree of positive feelings toward their ethnic group). The MEIM, one of the few measures available to assess ethnic identity development, has demonstrated strong reliability and has been widely used in the field of ethnic identity research. Although the measure is comprised of three subscales, it is typically used as a single scale and a summation of the 14 items is used to assess individuals degree of ethnic identity achievement. Thus, with its current use, the measure assesses ethnic identity achievement, in part, by calculating individuals responses to positive feelings about their ethnic group (e.g., I am happy that I am a member of the group I belong to ). In other words, an achieved ethnic identity applies to individuals who have made a commitment to their ethnic identity that is characterized by a period of personal exploration and positive feelings toward their ethnic group. Past research on ethnic identity has been conducted under this assumption that an achieved ethnic identity implies a positive identification with the group. In such research, the critical component has centered around one s positive response to one s ethnic group, rather than focusing on the process (e.g., ways in which individuals have explored their identity and developed an understanding of how they feel about that group membership). Although Phinney s theoretical postulation does not assume a positive commitment to the group, the measurement tool based on that theoretical model does. Therefore, in using the MEIM, only individuals whose commitment to their ethnic identity is positive are characterized as having an achieved identity. Consequently, the measurement is incongruent with the theory, as one s commitment is confounded with one s affirmation of one s ethnic identity. Furthermore, with current operationalizations of ethnic identity, it is difficult to understand the relation between ethnic identity and various outcomes. For exam-

5 ETHNIC IDENTITY SCALE 13 ple, based on existing work with the MEIM, findings suggest that it may be unhealthy to have a low degree of identification with one s ethnic group (e.g., lower self-esteem). With the current operationalization of the construct, however, it is not possible to decipher which aspect of the ethnic identity formation process is associated with negative outcomes. With current methods, exploration, commitment, and affirmation toward one s ethnicity are examined jointly. The key to furthering our understanding of the aspects of ethnic identity formation that are associated with outcomes may lie in examining exploration, commitment, and affirmation as distinct components of ethnic identity as opposed to using a sum score of a measure that combines individuals scores on the three constructs. We argue that current methods of assessing ethnic identity development are not entirely consistent with Erikson s original formulation of exploration and commitment. With current methods, identity resolution is examined along a continuum of exploration, commitment, and affirmation. Such a continuum precludes the possibility of creating statuses of ethnic identity by assuming a continuous process of identity negotiation that culminates in a positive assessment of ethnic group membership. Whereas Erikson s original conceptualization of commitment implied a resolution of how various component identities are related to the broader self, current definitions of commitment imply a positive assessment of the meaning of group membership despite the fact that identity achievement is defined as the secure sense of self that is the optimal outcome of the identity formation process (Phinney, 1992, p. 160). Although in theory such a resolution does not imply a positive or negative commitment, in practice, ethnic identity achievement has been assessed through questions measuring affirmation and belonging (e.g., I have a lot of pride in my ethnic group and its accomplishments ), whereby positive ratings imply a greater degree of ethnic identity achievement. Thus, we propose a new typology for examining ethnic identity statuses that is consistent with Marcia s operationalization of Erikson s theory and also consistent with Tajfel s social identity theory. Our typology mirrors Marcia s, by examining whether individuals degree of exploration and resolution regarding ethnicity is high or low. Consistent with Marcia s framework, our typology uses the statuses diffuse, foreclosed, moratorium, and achieved. Furthermore, our typology adds a third dimension that is in line with social identity theory. Based on individuals scores on affirmation, a positive or negative label is assigned to their diffuse, foreclosed, moratorium, or achieved status. Thus, an individual who scores low on exploration and commitment but high on affirmation would be categorized as diffuse positive, whereas an individual who scores low on all three components would be categorized as diffuse negative. By developing a measure that independently assesses the three distinct components of ethnic identity formation and thereby allows for the classification of individuals into an ethnic identity typology, the framework through which ethnic identity is examined can be refined and can more clearly capture its variability. In other words, the proposed typology would capture the experiences of individuals who feel that their ethnicity is an important component of their social selves, en-

6 14 UMAÑA-TAYLOR, YAZEDJIAN, BÁMACA-GÓMEZ gage in a process of exploration, resolve their feelings, and choose to affirm the role that their ethnic identity plays in their lives. Additionally, it would also capture the experiences of individuals who have explored their ethnicity and maintain a clear sense of what that group membership means to them, yet may not ascribe positive feelings toward their ethnic group. Take, for example, two Filipino women. They have both explored their ethnicity by attending cultural events, reading books about the history of the Philippines, and talking to their families about Filipino culture (i.e., exploration). In addition, they both feel confident about what being Filipino means to them (i.e., resolution). However, one of the women feels very positively about her Filipino background while the other woman feels negatively because of the history of colonization of the Philippines by multiple countries, which she feels has resulted in a lack of a unique Filipino culture (i.e., affirmation). Both women have explored and resolved the meaning of their group identification. Although they have reached divergent conclusions, they have nevertheless engaged in the same process. Thus, not only will it be important to examine the scales distinctively, but also interactively. Based on existing theoretical and empirical work on identity formation and, more specifically, ethnic identity formation, it is our premise that ethnic identity is comprised of three distinct components: (a) the degree to which individuals have explored their ethnicity, (b) the degree to which they have resolved what their ethnic identity means to them, and (c) the affect (positive or negative) that they associate with that resolution. Furthermore, we propose that these three components must be measured independently of one another to more accurately understand individuals statuses within the ethnic identity formation process and to be able to relate those statuses to important social and psychological outcomes. GENERAL METHOD Our primary objective was to develop a measure that assessed the three components of ethnic identity and, more specifically, to develop a measure that distinguished affirmation, resolution, and exploration. We initially devised the Ethnic Identity Scale (EIS) with only two subscales: exploration and resolution. Marcia s typology of identity achievement guided the development of the exploration and resolution items. The items assessed the degree to which individuals (a) had explored their ethnicity and (b) felt that they had resolved issues regarding their ethnicity. Following the development of the items, the authors invited four scholars to participate in a focus group to discuss the face validity of the items. During the discussion, items were reworded, deleted, and added to the original set of 35 items. Furthermore, it was determined that a third subscale should be included in the measure, and that this subscale should assess the affect that is ascribed to one s ethnicity. This idea emerged as a result of a discussion about how individuals may feel committed to an ethnic identity, yet that commitment may be based on negative or positive feelings toward the ethnicity. Following the focus group, the authors refined the existing

7 ETHNIC IDENTITY SCALE 15 items and created a new set of items to assess the third dimension affirmation. The development of items for the affirmation subscale was guided by Phinney s (1989) work on ethnic identity formation. The items of the scale were developed as definitive statements and answer choices were designed to reflect the degree of agreement or disagreement with the statements. In total, the questionnaire included 46 items (i.e., 23 exploration, 13 resolution, and 10 affirmation) and each subscale included both positively and negatively worded items. To examine the psychometric properties of the measure, we conducted two studies. The first study utilized an undergraduate sample and consisted of both exploratory and confirmatory factor analyses to determine the factor structure of the scale. In addition, the psychometric properties of the refined measure were examined in this study. The second study examined the methodological properties of the three subscales that were developed in Study 1 with a sample of high school students to determine whether the measure was valid and reliable with a younger adolescent population. STUDY 1 The purpose of our first study was to examine and refine the factor structure of the EIS. In addition, we explored the reliability and validity of the measure and examined whether individuals could be classified into the proposed ethnic identity typology based on their scores on the three components of ethnic identity. Finally, we examined whether individuals scores on self-esteem and familial ethnic socialization varied based on their typology classification. Method Sample and Procedure Data were gathered from 615 individuals attending either a 4-year university located in the Midwest (n = 297) or a 4-year university located on the West Coast (n = 318). The total sample included 164 men (27%) and 445 women (72%); six respondents did not provide this information. Participants ranged in age from 18 to 56 years (M = 21.8, SD = 3.91) and, in total, reported 193 different ethnic backgrounds (e.g., Poland, Mexican, Irish, and Eritrean). We used respondents self-reported ethnicity to classify them into six groups: White (n = 276, 45%), Latino (n = 71, 12%), Asian (for the purposes of this article, Asian includes peoples from East Asia, Southeast Asia, the Pacific Islands, South Asia, West Asia, and the Middle East; n = 125, 20%), Black (n = 49, 8%), multi-ethnic/racial (n = 41, 7%; e.g., Asian and White), and other (n = 25, 4%; e.g., American). Twenty eight participants (5%) did not report an ethnic background. Self-report questionnaires were administered in four upper-division, general education classes (e.g., Marriage and the Family, Interpersonal Communication) and

8 16 UMAÑA-TAYLOR, YAZEDJIAN, BÁMACA-GÓMEZ two lower-division introductory classes (i.e., Communication, Human Development), all of which draw students from majors across campus. A researcher explained the purpose of the project to the students and asked them to read the instructions and complete the questionnaire. Participation was voluntary. Questionnaires were administered on campus and took approximately 20 min to complete. Measures The questionnaire assessed various demographic variables (e.g., age, school year, ethnicity) and included measures that assessed ethnic identity formation, self-esteem, and familial ethnic socialization. EIS. A 46-item scale was developed to examine participants ethnic identity and, specifically, to assess three distinct aspects of ethnic identity: (a) exploration, (b) resolution, and (c) affirmation. The exploration subscale was comprised of 23 items (e.g., I have attended events that have helped me learn more about my ethnicity ); the resolution subscale included 13 items (e.g., I understand how I feel about my ethnicity ); and the affirmation subscale was comprised of 10 items (e.g., I have positive feelings toward my ethnicity ). Items for all subscales were scored on a 4-point Likert scale ranging from 1 (does not describe me at all)to4 (describes me very well). Responses were coded such that higher scores on each subscale indicated greater exploration, resolution, and affirmation. Race/Ethnicity. The following introduction and brief question preceded the EIS items in the questionnaire The U.S. is made up of people of various ethnicities. Ethnicity refers to cultural traditions, beliefs, and behaviors that are passed down through generations. Some examples of the ethnicities that people may identify with are Mexican, Cuban, Nicaraguan, Chinese, Taiwanese, Filipino, Jamaican, African American, Haitian, Italian, Irish, and German. In addition, some people may identify with more than one ethnicity. When you are answering the following questions, we d like you to think about what YOU consider your ethnicity to be. Please write what you consider to be your ethnicity here and refer to this ethnicity as you answer the questions below. Participants responses to this question were then used to create the variable race/ethnicity. Self-esteem. Rosenberg s (1979) Self-esteem Scale was used to assess participants global self-esteem. This measure is comprised of 10 items (e.g., At times I think I am no good at all ) with a scale ranging from 1 (strongly disagree) to4 (strongly agree). Items were scored such that higher scores indicated higher self-esteem. This scale has been used with ethnically diverse populations (e.g., Mexican, Dominican, Puerto Rican, African American, and White adolescents) and has obtained moderate coefficient alphas (e.g.,.79 to.85) with these samples (Der-Karabetian & Ruiz, 1997; Lorenzo-Hernandez & Ouellette, 1998; Martinez

9 ETHNIC IDENTITY SCALE 17 & Dukes, 1997; Phinney, Cantu, & Kurtz, 1997). With the current sample, the scale obtained a coefficient alpha of.85. Familial Ethnic Socialization. A revised version of the Familial Ethnic Socialization Measure (FESM; Umaña-Taylor, 2001) was used to assess the degree to which participants perceived that their families socialized them with respect to their ethnicity. The 12 items (e.g., My family teaches me about our family s ethnic/cultural background and Our home is decorated with things that reflect my ethnic/cultural background ) were rated on a 5-point Likert scale ranging from 1 (not at all true)to5(very much)and1(not at all)to5(very often). Responses were coded so that higher scores indicated higher levels of familial ethnic socialization. The original version, consisting of 9 items, obtained a coefficient alpha of.82 with a sample of Mexican-origin adolescents (Umaña-Taylor, 2001). The revised scale obtained a coefficient alpha of.94 with the current sample. Results and Discussion As previously mentioned, Study 1 involved two steps. Because we sought to examine the factor structure and psychometric properties of the 46-item measure with as diverse a sample as possible (i.e., diverse with regard to ethnicity and geographical locale), we randomly selected half of the sample from a university on the West coast and half of the sample from a university in the Midwest. We then combined these two randomly selected participant groups into the sample for the first step of Study 1. With samples large enough to split randomly in half (as was the case in this study), it is suggested that exploratory factor analysis can be used to discover a factor structure and confirmatory factor analysis can be used to refine and confirm the basic model using the second half of the sample (Bryant & Yarnold, 1995). As such, the first step in Study 1 involved conducting an exploratory factor analysis using maximum likelihood extraction and an oblique rotation with the original 46 items. Eight factors met the Kaiser (1958) retention criterion of eigenvalues greater than 1 and accounted for 66% of the variance. After careful examination of the factor loadings and scree plot, a three-factor solution, which explained 49% of the variance, was retained. That is, items were retained if they loaded higher than.40 on one of the three factors that explained the most variance and loaded exclusively on one factor. This solution was conceptually consistent with our theoretical framework and was psychometrically robust. This resulted in a 22-item measure, comprised of three subscales that were named exploration, affirmation, and resolution. The subscales were derived directly from the three-factor solution and consisted of seven, six, and nine items, respectively (see Table 1 for eigenvalues, loadings of variables on factors, and percentages of variance). The internal consistency of the exploration, affirmation, and resolution subscales was examined and alpha coefficients were strong:.90,.86, and.86, respectively. The combination of moderately high coefficient alphas

10 18 UMAÑA-TAYLOR, YAZEDJIAN, BÁMACA-GÓMEZ TABLE 1 Exploratory Factor Analysis of the Ethnic Identity Scale (EIS) Study 1 Factor 1 a Factor 2 b Factor 3 c I have participated in activities that have taught me about my ethnicity (EIS 39) I have participated in activities that have exposed me to my ethnicity (EIS 19) I have read books/magazines/newspapers or other materials that have taught me about my ethnicity (EIS 15) I have attended events that have helped me learn more about my ethnicity (EIS 13) I have learned about my ethnicity by doing things such as reading (books, magazines, newspapers), searching the Internet, or keeping up with current events (EIS 26) I have not participated in any activities that would teach me about my ethnicity (EIS 7) I have experienced things that reflect my ethnicity, such as eating food, listening to music, and watching movies (EIS 11) If I could choose, I would prefer to be of a different ethnicity (EIS 36) I wish I were of a different ethnicity (EIS 20) I feel negatively about my ethnicity (EIS 16) I dislike my ethnicity (EIS 40) I am not happy with my ethnicity (EIS 24) My feelings about my ethnicity are mostly negative (EIS 2) I am still trying to understand what my ethnicity means to me (EIS 32) I am still trying to understand how I feel about my ethnicity (EIS 18) I am not clear about my feelings about my ethnicity (EIS 34) I am not clear about how I feel about my ethnicity (EIS 46) I am not sure how I feel about my ethnicity (EIS 22) I am clear about what my ethnicity means to me (EIS 10) I understand how I feel about my ethnicity (EIS 29) I have a clear sense of what my ethnicity means to me (EIS 42) I know what my ethnicity means to me (EIS 38) Eigenvalue % variance % cumulative variance Coefficient alpha Note. N = 250. Bolded numerals denote factor loadings of.40 or higher. a b c Exploration subscale. Affirmation subscale. Resolution subscale.

11 ETHNIC IDENTITY SCALE 19 and the distinct factor structure that emerged in the exploratory factor analysis provided empirical support for the three components of ethnic identity that had been theoretically generated. Our next step involved conducting a confirmatory factor analysis, using maximum likelihood, to refine and confirm the basic model. The sample for this analysis included the remaining half of each university sample. Initially, a three-factor model was examined, which included the 22 observed variables (see Figure 1). In the figure, observed variables are enclosed in rectangles and latent variables, also known as factor variables, are enclosed in ovals. Examination of the fit indexes in- FIGURE 1. Confirmatory Factor Analysis for initial 22-item model

12 20 UMAÑA-TAYLOR, YAZEDJIAN, BÁMACA-GÓMEZ dicated that the model was a poor fit to the data (Goodness of Fit Index, GFI =.68, Comparative Fit Index, CFI =.79, χ 2 /df = 5.12, Root Mean Square Error of Approximation, RMSEA =.12). In sum, the combination of indexes suggested that the model did not adequately represent the relations exhibited within the data. As a result, the standardized residual matrix was examined to identify potential problems with the model (McDonald, 1999). Examination of the residual matrix indicated that five items had residuals that were significantly large (p <.05). These five items, all from the resolution subscale, were discarded and a second model, containing 17 items, was examined (see Figure 2). FIGURE 2. Confirmatory Factor Analysis for final 17-item model

13 The second model demonstrated a significant improvement in fit, as evidenced by a significant change in the χ 2 value (Model 1 χ 2 (206, N = 269) = ; Model 2 χ 2 (116, N = 269) = ; difference in χ 2 = , difference in df = 90, p <.001) and a lower value on the Akaike s Information Criterion (AIC; Model 1 AIC = , Model 2 AIC = ). The AIC can be utilized as a measure of model comparison, similar to the adjusted R 2 in multiple regression. Lower AIC values indicate a more parsimonious and better fitting model. In addition, multiple fit indexes indicated adequate model fit (GFI =.86, CFI =.91, RMSEA =.09). Reliability To determine the internal consistency of each subscale, coefficient alphas were examined. Coefficient alphas for the exploration, affirmation, and resolution subscales were high (.91,.86, and.92, respectively). Validity ETHNIC IDENTITY SCALE 21 To assess the construct validity of the EIS, correlations among the three subscales (i.e., exploration, affirmation, and resolution), a measure of self-esteem, and a measure of familial ethnic socialization were examined. Because previous work demonstrates that the salience of ethnic identity varies by group membership, we decided to examine these correlations separately for majority group members (i.e., Whites) and ethnic minority group members. First, we examined the intercorrelations among the subscales of the EIS. Findings indicated that the exploration and resolution subscales were strongly correlated with one another for both White students and ethnic minority students (see Table 2). Thus, individuals who reported high levels of exploration regarding their ethnicity also tended to report high levels of resolution. Due to the correlational nature of the analyses, it is not possible to determine whether exploration influences one s resolution or whether individuals who have resolved how they feel about their ethnicity are more apt to explore their ethnicity at greater lengths. It is quite possible that the relation is bidirectional. Theoretically, one would expect that through exploration one can come to a resolution; however, it is possible that after one has resolved issues related to ethnicity, one will continue to explore and seek out activities that will expose one to one s ethnicity. With regard to affirmation, this subscale was significantly related to the exploration and resolution subscales for ethnic minority group members and there was a trend in this direction for Whites. Although a significant correlation emerged, this relation was not as strong as the relation that emerged between exploration and resolution, suggesting that exploration and/or resolution regarding one s ethnicity is not necessarily related to the affect that one holds toward that ethnicity. Put differently, we cannot assume that individuals will feel positively about their ethnicity just because they have explored their ethnicity and/or feel that they have resolved

14 22 TABLE 2 Pearson Correlations among the Ethnic Identity Scale (EIS) subscales, the Rosenberg Self-Esteem Scale, and the Umaña-Taylor Familial Ethnic Socialization Measure for Study 1 and Study 2 By Ethnic Group Study 1: University Study 2: High School M SD M SD White students 1. Exploration Affirmation Resolution.58*** * Self-esteem *.37** Familial ethnic socialization.64***.06.62*** ***.11.49*** Ethnic minority students 1. Exploration Affirmation.23** * Resolution.60***.22** ***.18* Self-esteem.27**.10.36*** **.21*.27** Familial ethnic.61***.08.41***.18* ***.03.44*** socialization + <.10. *p <.05. **p <.01. ***p <.001.

15 ETHNIC IDENTITY SCALE 23 how they feel about their ethnicity. This finding provides limited support for our premise that individuals commitment toward their ethnicity may be accompanied by either positive or negative affect and should be explored further. To explore the construct validity of the measure, we examined the correlations of each subscale and measures of self-esteem and familial ethnic socialization. Previous research has found significant positive relations between ethnic identity and self-esteem (for reviews see Phinney, 1995; Umaña-Taylor et al., 2002), as well as between ethnic identity and familial ethnic socialization (Umaña-Taylor & Fine, 2001). In this study, the exploration and resolution subscales were both positively associated with self-esteem and familial ethnic socialization among ethnic minority group members. For majority group members, there was a significant relation between the exploration and resolution subscales and familial ethnic socialization but not for self-esteem, although that relation approached significance. For both groups, however, respondents scores on the affirmation subscale were not significantly correlated with their scores on self-esteem or familial ethnic socialization. In previous work, researchers have found a positive, yet modest, relation between ethnic identity and self-esteem. An important note, however, is that previous studies have used the MEIM, which combines individuals scores on the three constructs we have measured in this study. It is possible that the relation that emerged between ethnic identity and self-esteem in previous studies was only modest because the method for measuring ethnic identity introduced error by combining individuals scores on affirmation with their scores of exploration and commitment. In other words, perhaps the fact that affirmation was always considered a part of ethnic identity could explain why findings were not as strong. Our findings suggest that it is not how positively or negatively you feel about your ethnicity that influences or is related to your self-esteem; rather, it is how much you feel that you have explored or resolved issues relating to your ethnicity. With regard to familial ethnic socialization, individuals scores on exploration and resolution were positively related to their scores on familial ethnic socialization. Consistent with previous research (i.e., Umaña-Taylor & Fine, 2001, 2002), individuals who reported that their families had socialized them with regard to their ethnicity also tended to report higher levels of exploration and resolution regarding their ethnicity. Interestingly, individuals scores on the affirmation subscale were not related to their scores on familial ethnic socialization. Again, because previous research in which familial ethnic socialization and ethnic identity have been examined has used the MEIM to measure ethnic identity, it is unclear whether individuals scores on exploration and commitment have driven this relation. Nevertheless, our findings suggest that individuals experiences with familial ethnic socialization are not related to the affect that they report toward their ethnicity. Distribution of EIS Scale Scores. Because we were interested in developing a measure that could be used to categorize individuals into a typology, we set out to

16 24 UMAÑA-TAYLOR, YAZEDJIAN, BÁMACA-GÓMEZ identify cutoff values for each of the subscales, whereby individuals could be classified into eight groups based on their scores on each of the three subscales (i.e., all possible combinations of high/low affirmation, high/low exploration, high/low resolution). Specifically, we were interested in examining whether individuals could be classified in each of the eight types and if their scores on outcome variables (i.e., familial ethnic socialization and self-esteem) would differ as a result of their group classification. However, it was unclear what cutoffs should be used to define a high or a low score. A variation of a K-means Cluster Analysis was used to find the proposed types. The K-means Cluster Analysis searches for the best eight groups from the three subscales such that the Sum of Squares Within (SSW) is minimized. (The SSW is defined as the sum across all observations of squared Euclidean distance from the observation to the centroid [a point that represents the average score of the three subscales] for each group.) Because our goal was to find the best groups based on the restriction that groups must be derived from the high-low scoring of each subscale, it was necessary to modify the search from the K-means Cluster algorithm, such that the only possible groupings to consider would be derived using the high-low scores from cutoffs of each subscale (Anderberg, 1973; Späth, 1980). The cutoffs that resulted in groups in which the SSW was the lowest were used as the best cutoffs for the data. The cutoff values that resulted were 19.5, 20.5, and 9.5 for the exploration, affirmation, and resolution subscales, respectively. To better contextualize these figures, response choices were measured on a scale ranging from 1 (does not describe me at all)to4 (describes me very well). Because sum scores were utilized, respondents scores could range from 7 to 28 on the 7-item exploration subscale, 6 to 24 on the 6-item affirmation subscale, and 4 to 16 on the 4-item resolution subscale. After the cutoff values were determined, participants scores were categorized into eight categories. Respondents who scored above the cutoff value on a particular subscale were considered to score high on that subscale and those who scored at or below the cutoff value were considered to score low on the subscale. The eight types were labeled diffuse negative, diffuse positive, foreclosed negative, foreclosed positive, moratorium negative, moratorium positive, achieved negative, and achieved positive (see Table 3). For example, the label achieved positive identifies individuals who scored above the cutoff value on all three subscales (i.e., achieved = high exploration and high commitment, positive = high affirmation). Based on this categorization scheme, we were able to classify all individuals into one of the eight types. The distribution of scores indicated that only 10% of the sample reported low scores on the affirmation subscale. Furthermore, the group with the largest number of people included those who had explored, had resolved issues about their ethnicity, and felt positively about their ethnicity (i.e., 38% of the sample). In line with this, previous research shows a general trend for individuals to increasingly move toward ethnic identity achievement in young adulthood (Phinney, 1992; Phinney & Chavira, 1992). Nevertheless, it is possible that the distribution of scores was

17 ETHNIC IDENTITY SCALE 25 TABLE 3 Distribution Of Scores Along High/Low Dimensions of Exploration, Resolution, and Affirmation Subscales for Study 1 and Study 2 Study 1: University Study 2: High School Affirmation Affirmation Low High Low High n % n % n % n % Diffuse (low exploration, low resolution) Foreclose (low Exploration, high resolution) Moratorium (high exploration, low resolution) Achieved (high exploration, high resolution) driven, in part, by social desirability. It is possible that individuals were inclined to report positive affect toward their ethnicity (i.e., affirmation) and that is why only 10% of our sample fell into the negative categorization. The next two largest groups were (a) those who had not explored or resolved and felt positively about their ethnicity (i.e., 30%) and (b) those who had not explored but reported that they had resolved issues regarding their ethnicity and that they felt positively about it (i.e., 18%). This is an important finding because it demonstrates that it is possible to distinguish individuals who are diffuse or foreclosed. Thus, it may be inaccurate to have a combined category of diffuse and foreclosed because they involve different processes (although both involve low exploration, one refers to individuals who have resolved issues regarding their ethnicity and the second involves those who feel that they have not resolved issues about their ethnicity). A cautionary note, however, is that the resolution and exploration subscales are significantly correlated with one another and this relation could produce an overlap in the status categories that are derived from these subscale scores. Future studies, with larger samples, should explore the distribution of individuals typology classification. To explore the distribution of typology classifications with regard to ethnic minority/majority status, we examined the number of minority and majority individuals who fell into each of the eight groups. Because previous research has demonstrated that ethnic identity has different meanings for ethnic minority and ethnic majority individuals, we were interested in exploring whether one group would dominate certain classifications. Examination of the distribution suggests somewhat distinct patterns for ethnic minority and ethnic majority individuals (see Table 4). The majority of ethnic majority individuals (i.e., White) tend to fall

18 TABLE 4 Study 1 Distribution of Ethnic Groups by Ethnic Identity Classification Ethnic Identity Classification Panethnic Group White Latino Asian Black Multi-ethnic/racial Other (n) (%) (n) (%) (n) (%) (n) (%) (n) (%) (n) (%) Diffuse negative Diffuse positive Foreclosed negative Foreclosed positive Moratorium negative Moratorium positive Achieved negative Achieved positive Total

19 ETHNIC IDENTITY SCALE 27 into the Diffuse Positive category, which is associated with low exploration, low commitment, but positive feelings about one s ethnic group membership. On the other hand, the majority of Latino, Asian, and Black individuals tend to fall into the Achieved Positive category, indicating that they have explored, committed, and feel positively about their ethnic group membership. These patterns are merely presented as descriptive data; however, they support previous findings that indicate ethnic identity is a more salient construct for ethnic minority group members than for ethnic majority group members (Phinney, 1989, 1992; Phinney & Alipuria, 1990; Roberts et al., 1999). Mean Differences Between Types. In addition to examining whether individuals could be classified into the typology, we were interested in examining whether individuals scores varied on other variables, based on their typology classification. Specifically, we conducted two one-way analyses of variance (ANOVAs) to determine whether individuals familial ethnic socialization and self-esteem varied as a function of their typology classification. Results indicated that individuals scores on familial ethnic socialization varied significantly depending on EIS type, F(7, 277) = 32.23, p <.001. On the contrary, individuals self-esteem scores did not vary based on EIS type, F(7, 275) = 1.67, ns. To follow up the significant main effect for familial ethnic socialization, we conducted Tukey s post hoc analyses (see Table 5 for means and standard deviations). Results indicated that (a) individuals who were classified as achieved positive reported significantly higher levels of familial ethnic socialization than TABLE 5 Study 1 Means And Standard Deviations Of Familial Ethnic Socialization and Self-Esteem by Ethnic Identity Scale (EIS) Type Familial Ethnic Socialization Self-Esteem n M SD n M SD Diffuse negative c Diffuse positive bdefgh Foreclosed negative h Foreclosed positive ad Moratorium negative g Moratorium positive e Achieved negative f Achieved positive abc F(df) 32.23*** (7, 277) 1.67 (7, 275) Note. Values in same column with same superscript are significantly different from one another. * p <.05. ** p <.01. *** p <.001.

20 28 UMAÑA-TAYLOR, YAZEDJIAN, BÁMACA-GÓMEZ individuals classified as foreclosed positive, diffuse positive, and diffuse negative (p <.001 for all) and (b) individuals classified as diffuse positive reported significantly lower levels of familial ethnic socialization than individuals classified as foreclosed negative, moratorium negative, moratorium positive, and achieved negative (p <.01 for all). Thus, adolescents who reported high levels of exploration, resolution, and affirmation (i.e., achieved positive) reported higher levels of familial ethnic socialization than those who reported feeling positively but not necessarily exploring (i.e., foreclosed positive) or resolving (i.e., diffuse positive). These findings provide further support for the construct validity of the subscales because individuals who report low exploration and/or resolution also tend to report low levels of familial ethnic socialization. For self-esteem, findings were not significant but further examination of the means demonstrates a trend in that those who have explored, resolved, and feel positively tend to have the highest self-esteem scores. These findings are consistent with Tajfel s and Erikson s theoretical perspectives. Finally, although we were unable to explore differences within ethnic groups by typology classifications because of limited sample size, we recommend that future studies explore the possible differences by including large samples of ethnic group members. STUDY 2 The purpose of our second study was to examine if the measure that was developed with a university sample would be applicable to a sample of high school students. Thus, in Study 2 we examined the reliability and validity of the EIS with 231 eleventh graders. In addition, we examined the distribution of scores in the typology and compared scores on self-esteem and familial ethnic socialization among groups. Method Sample and Procedure Data for this study were drawn from a larger research project examining the influence of school context on ethnic identity. Data were gathered from adolescents attending a high school located in a large ethnically diverse city in the Midwest. The sample included 231 eleventh grade high school students (45.5% male students, 54.5% female students). Age of participants ranged from 15 to 18 (M = 16.6, SD =.59). Participants reported 88 different ethnic backgrounds (e.g., Polish, Vietnamese, Mexican, Native American) from which pan-ethnicclassifications were created. Of all participants, 64 (28%) were White, 49 (21%) were Latino, 35 (11%) were Asian, 46 (20%) were Black, 3 (1%) were Native American, 18 (8%) were multi-ethnic/racial, and 6 (3%) were classified as other. A total of 10 participants (4%) did not report their ethnic/racial background.

21 Self-reported questionnaires were administered in English classes and took approximately 25 min to complete. Participants were informed that their participation in the study was voluntary and the purpose of the research project was explained before students completed the questionnaire. A researcher was present to answer any questions or to clarify terms when needed. Measures ETHNIC IDENTITY SCALE 29 The questionnaire assessed various demographic factors (e.g., age, school year, ethnicity, family information) and school-related factors and included measures of self-esteem, familial ethnic socialization, ethnic identity formation, school climate, and social networks. For the purposes of this study, only the demographic information and the measures of self-esteem, familial ethnic socialization, and ethnic identity formation were examined. A detailed description of the self-esteem and familial ethnic socialization measures is provided in Study 1. With this sample, coefficient alphas for the self-esteem scale and the FESM were.88 and.92, respectively. The final version of the EIS that emerged from Study 1 was used in this study to examine participants ethnic identity. This measure included 17 items that assessed the degree to which individuals (a) have explored aspects related to their ethnicity, (b) have resolved any issues related to their ethnicity, and (c) feel positively about their ethnicity. The final version of the EIS was composed of seven items that assessed exploration, four items that assessed resolution, and six items that assessed affirmation (see Appendix for complete scale). Items were scored so that higher scores indicated higher levels of exploration, resolution, and affirmation. Results and Discussion Reliability Alpha coefficients were examined to determine whether the strong internal consistency that had been observed among the university students would be replicated with a sample of high school students. In line with results from the university sample, coefficient alphas among the high school sample were moderately high for the exploration, affirmation, and resolution subscales (.89,.84,.89, respectively). Validity Similar to the procedure we followed in Study 1, the intercorrelations among the subscales of the EIS were examined separately by ethnic majority/minority group. In line with findings from Study 1, the exploration and resolution subscales were significantly positively correlated with one another for both groups. With the high school sample, however, a significant positive relation emerged between ethnic minority adolescents scores on the affirmation and exploration subscales, whereas this relation did not emerge among the White adolescents. Thus, ethnic