Journal of Personality and Social Psychology 1998, Vol. 74, No. 3, 715-724 Copyright 1998 by the American Psychological Association, Inc. 0022-3514/98/53.00 The Relationship Between Racial Identity and Self-Esteem in African American College and High School Students Stephanie J. Rowley, Robert M. Sellers, Tabbye M. Chavous, and Mia A. Smith University of Virginia The Multidimensional Model of Racial Identity was used to examine the relationship between racial identity and personal self-esteem (PSE) in a sample of African American college students (n = 173) and a sample of African American high school students (n = 72). Racial identity was assessed using the Centrality and Regard scales of the Multidimensional Inventory of Black Identity, whereas the Rosenberg Self-Esteem Scale was used to assess PSE. Riur predictions were tested: (a) racial centrality is weakly but positively related to PSE (b) private regard is moderately related to PSE (c) public regard is unrelated to PSE and (d) racial centrality moderates the relationship between private regard and PSE. Multiple regression analysis found that racial centrality and public racial regard were unrelated to PSE in both samples. Private regard was positively related to PSE in the college sample. Racial centrality moderated the relationship between private regard and PSE in both samples, such that the relationship was significant for those with high levels of centrality but nonsignificant for those with low levels. At the beginning of this century, W. E. B. Dubois argued that the central problem of American society was that of the color line (Dubois, 1903). An examination of the psychological litera- ture over the past 90 years suggests that Dubois was prophetic. Social psychology has been exploring the meaning of race in the lives of African Americans since its earliest inception. Inter- estingly, much of the research produced by mainstream psychol- ogy suggests that African Americans who identify strongly with being Black may be at psychological risk as a result of the stigma associated with the identity (e.g., Horowitz, 1939 Kardiner & Ovesey, 1951 Steele & Aronson, 1995). Other researchers have tended to suggest that a strong identification with one's race can serve as a protective buffer to personal self-esteem (e.g., Azibo, 1989 Baldwin, 1984 Cross, 1971, 1991). These two approaches suggest a conundrum with respect to the "true" relationship between racial identification and personal self-es- teem in African Americans. In this article, we attempt to explain this conundrum using the Multidimensional Model of Racial Identity (Sellers et al., in press Sellers, Smith, Shelton, Row- ley, & Chavous, 1998). African American Racial Identity and Self-Esteem Research Much of the early research in racial identity assumed that African Americans have negative self-images (Cross, 1991). It Stephanie J. Rowley, Robert M. Sellers, Tabbye M. Chavous, and Mia A. Smith, Department of Psychology, University of Virginia. Stephanie J. Rowley is now at Department of Psychology, University of North Carolina at Chapel Hill, Correspondence concerning this article should be addressed to Robert M. Sellers, who is now at Department of Psychology, 3253 East Hall, University of Michigan, Ann Arbor, Michigan 48109. Electronic mail may be sent to rsellers@umich.edu. was assumed that because one's self-concept is influenced by perceptions of the way one is viewed by others, society's nega- tive views of African Americans must be internalized (Kar- diner 3c Ovessey, 1951 Pettigrew, 1978). Thus, the prevailing opinion among psychologists at that time was that African Americans suffered from an inferiority complex, and conse- quently, low self-esteem (Cross, 1991). Findings such as those reported by Clark and Clark (1947), in which African American children demonstrated a preference for playing with White dolls, were interpreted as evidence that African American children hate themselves for being Black and wish that they were White. Over a half century ago, Clark and Clark (1940) noted the problem of confounding racial identification with racial prefer- ence in the work of Ruth Horowitz (cf. Cross, 1991). Unfortu- nately, few researchers have adequately heeded this warning. The assumption that African Americans suffer from an inferior- ity complex (i.e., low self-esteem) went virtually unchallenged until the late 1960s and early 1970s when researchers challenged it on conceptual and methodological grounds (e.g., Banks, 1976 Brand, Ruiz, & Padilla, 1974 McAdoo, 1970 Rosenberg & Simmons, 1971). Researchers astutely noted that the early work on African American racial identity erroneously assumed that reference group orientation (preference for same race stimuli) was related to, if not synonymous with, self-concept (e.g., Brand et al., 1974 McAdoo, 1970 Porter & Washington, 1979). In fact, research using measures of reference group orientation often was interpreted as if self-concept had been measured, although no explicit measure of self-concept was used (Cross, 1991). A preference for White stimuli was enough for research- ers to conclude that they had found support for self-hatred in the African American participants. When paper-and-pencil measures of self-esteem became more available and accepted in the late 1960s, researchers finally began to measure self-esteem in African Americans directly. Since then, numerous studies have concluded that African 715

716 ROWLEY, SELLERS, CHAVOUS, AND SMITH Americans do not suffer from low self-esteem (e.g., Hughes & Demo, 1989 Porter & Washington, 1979 Rosenberg, 1979b Rosenberg & Simmons, 1971 Taylor & Walsh, 1979). More recently, researchers have suggested that the inconsistencies be- tween reflective appraisal theory and the actual mental health of African Americans can be explained by the insulating effect of the African American community (Broman, Neighbors, & Jackson, 1989 Rosenberg, 1979b Rosenberg & Simmons, 1971). The insulation hypothesis argues that because of racial segregation in the United States, the majority of African Ameri- cans compare themselves not with members of the broader soci- ety, but with other African Americans. As a result, African Americans are insulated from the broader society's negative perceptions of their racial group, and their personal self-esteem is protected (Broman et al., 1989 Rosenberg & Simmons, 1971). One potential buffer against society's negative view to- ward African Americans is a healthy racial identity. A number of African American researchers have focused on the resilient strengths of African American experiences (Banks, 1970 Bur- lew & Smith, 1991 Cross, 1971 Kambon, 1992 Milliones, 1980). In contrast to earlier models that were based on the negative impact of oppression on African Americans, models of resilience argue that identification with one's race should result in more positive mental health outcomes such as high self- esteem (e.g., Baldwin, 1984). Thus, identification with one's race is conceptualized as a protective factor for personal self- esteem in African Americans. Despite these assertions, research investigating the relation- ship between racial identity and self-esteem is far from conclu- sive (Brand et al., 1974 Hughes & Demo, 1989 Porter & Wash- ington, 1979). In a review of the African American racial iden- tity and self concept literature, Cross (1991) examined the results from 45 studies conducted from 1939 to 1987. Cross found that 36% of the studies reported a significant positive association between racial identity and self-esteem, whereas 64% of the studies reported no relationship. The majority of the 45 studies used children and teens as participants (n = 34). Of the 11 studies that focused on adults, only 3 found results that suggested a positive relationship. One reason for the inconclusiveness of the literature is that racial identity researchers have failed to explicate the mecha- nism by which a strong racial identity should result in higher levels of self-esteem (Penn, Gaines, & Phillips, 1993 Sellers, 1993). This gap in the literature is, in part, a function of the fact that many models of racial identity fail to conceptually define dimensions of racial identity that relate to self-esteem. Many existing racial identity measures include items that mea- sure something akin to racial self-esteem (e.g., Baldwin, Dun- can, & Bell, 1987 Parham & Helms, 1981). Unfortunately, the relationship between these items and personal self-esteem become obscured when investigators choose to include other aspects of racial identity that are unrelated to racial self-esteem in a single aggregate score. When only items that tap into af- fective and evaluative aspects of racial identity are included in the model, relationships between racial identity and personal self-esteem are apparent. Parham and Helms (1985), for in- stance, found that both pre-encounter attitudes (the strongest anti-Black attitudes in their stage model) and immersion-emer- sion attitudes (the strongest pro-Black attitudes) were negatively related to self-regard, whereas encounter (questioning initial anti-Black attitudes) attitudes were positively related to self- regard. The measure used to assess the stages of identity devel- opment, the Racial Identity Attitude Scale (Parham & Helms, 1985), however, is comprised of items that rely heavily on the individuals' affective and cognitive evaluations of African Americans (i.e., racial self-esteem). The few studies that have measured constructs similar to racial self-esteem separate from other dimensions of racial iden- tity have found a positive relationship between the constructs and personal self-esteem in African Americans. For instance, Hughes and Demo (1989) operationalized racial self-esteem as "the belief that most black people possess positive characteris- tics and do not possess negative characteristics" in a nationally representative sample of African Americans (p. 140). They found a small, but statistically significant, relationship between racial self-esteem and personal self-esteem such that individuals who reported more positive beliefs about African Americans had higher levels of personal self-esteem. Although it was not a major focus of their study, Crocker, Luhtanen, Blaine, and Broadnax (1994) reported evidence that African American col- lege students' scores on the private and membership subscales of the Collective Self-Esteem Scale (CSES) were positively correlated to personal self-esteem. Specifically, those African American students who felt more positive about African Ameri- cans and felt more positive about being African American had higher self-esteem scores. Interestingly, the African American students' perceptions about how others evaluated their race was not related to their own evaluations of their race nor their self- esteem scores. This finding is consistent with the insulation theory in that African American students' feelings about African Americans were more relevant to personal self-esteem than were their perceptions of how others view African Americans. An- other reason for the equivocal nature of the literature on racial identity and self-esteem is that many models of African Ameri- can racial identity presume that race is the most central aspect of the self-concept for all African Americans. As noted by the work of Ingram (1989) and Phinney and Alipuria (1990), such an assumption may not be tenable. Ingram (1989) used the Role Construct Repertory Grid to assess the meaningfulness of various self-constructs in a sample of African American college students. Seventy-two percent of her sample was female. She found that although participants rated race as a meaningful self- construct, gender was rated as the most meaningful construct. Phinney and Alipuria (1990) asked participants to rate five iden- tity domains (occupation, politics, religion, sex role, and eth- nicity) on a 4-point scale ranging from not at all important to very important. Their results revealed that ethnic identity tied for third place with religious identity among the African American participants in their sample. Cross (1991) used the term experimenter-ascribed identity to describe the situation where the experimenter makes assump- tions about participants' racial identity simply because they have physical characteristics that identify them as belonging to a particular racial group. In contrast, personally affirmed group identities are those groups that the individual uses to define himself or herself. The literature's dependence on experimenter- ascribed racial identity instead of personally affirmed identities has obscured the true relationship between racial identity and