Psychological Bulletin 21300, Vol. 126, No. 1, 26-54

Copyright 2000 by the American Psychological Association, Inc. 0033-2909/00/$5.00 DOI: 10.1037//0033-2909.126.1.26

Factors Influencing Racial Comparisons of Self-Esteem: A Quantitative Review Bernadette Gray-Little and Adam R. Hafdahl University of North Carolina at Chapel Hill

Research on racial comparisons of self-esteem was examined. Early research in this area, exemplified by the doll studies of racial preference, was viewed as demonstrating that Blacks have less self-regard than Whites. However, a meta-analytic synthesis of 261 comparisons, based largely on self-esteem scales and involving more than half a million respondents, revealed higher scores for Black than for White children, adolescents, and young adults. This analysis further revealed that the direction and magnitude of racial differences are influenced by such demographic characteristics as participant age and socioeconomic status, as well as by characteristics of the measuring instruments. Many findings--for example, that the self-esteem advantage for Black respondents increases with age and is related to the sex composition of the sample--underscore the need for long-term longitudinal studies of self-esteem development in male and female members of both racial groups.

The concept of self has been a focus of study in psychology for more than a century. For roughly half that time, substantial attention has been devoted to the relevance of race to self-concept and self-esteem. The self often has been described as a social product whose content depends on social interactions. Accordingly, it might be expected that membership in racial groups, as in other groups, would interact with a variety of perceptual and cognitive processes to influence individual self-construal and selfevaluation. The intense interest in race and self-esteem stemmed from the belief that racial comparisons (Black-White) offered a "natural" test of theories emphasizing the social basis of selfesteem in combination with the conviction that self-esteem played a crucial role in psychological well-being. The precise nature of the association between individual selfevaluation and race is often debated. However, most attention has focused on social approval and achievement as important explanatory mechanisms. For example, Cartwright (1950), theorizing from a reflected appraisal perspective, argued that Blacks and members of other oppressed or stigmatized groups would make low self-evaluations because they are held in low regard by others. Gerth and Mills (1953) premised that unfavorable social comparisons led members of social minorities to have lower self-esteem. They reasoned that "self-esteem is rather immediately a function of status position" (p. 325) and that the low status of minority group members was readily apparent in their exclusion from desirable occupations and social activities. Causal attribution theory is the basis for a more recent third theory of low self-esteem among Blacks, according to which persons of lower social status

are assumed to have fewer opportunities to control their environment, leading to reduced personal efficacy and lower self-esteem. (See Graham, 1994, for a review of research on personal efficacy in African Americans.) Whether generated by reflected appraisal, social comparison, or attribution theory, the hypothesis of low Black self-esteem has spawned scores of studies during more than 50 years of investigation (Gray-Little & Duley, 1993). Our goals in the present review were to synthesize this literature by comparing the average self-esteem levels of Black and White children, adolescents, and young adults, and to examine factors associated with the differences found. M. C. Taylor and Walsh (1979) noted that the expectation of low Black self-esteem was almost an axiom among social scientists until an accumulation of counterintuitive empirical findings during the 1970s forced a reconsideration of the association between race and self-esteem. Theories used to predict low selfesteem in Blacks were often recast to explain equal and high self-esteem among Blacks. For example, Rosenberg (1979) concluded that self-esteem levels of Black children and adolescents equaled those of Whites because their reflected appraisals and social comparisons stemmed from members of their own racial group, not White society. Hoelter (1982) explained the Black self-esteem advantage in terms of "selective credulity," a greater emphasis on external sources of affirmation and reduced reliance on internalized standards, as compared with Whites. (A similar formulation has been offered to explain the different bases for self-esteem in women and men, respectively. See S. 'E. Cross & Madson, 1997, and Josephs, Marcus, & Tafarodi, 1992.) Although Hoelter did find that Black adolescents stressed interpersonal relationships more than did Whites, controlling for this effect did not eliminate the Black self-esteem advantage, nor was there evidence of a closer link between self-esteem and internalized standards for White adolescents. M. C. Taylor and Walsh tested a complementary hypothesis--that work-related dimensions (a competency factor) would be more predictive of self-esteem in White than Black adults--but found opposite trends.

Bernadette Gray-Little and Adam R. Hafdahl, Department of Psychology, University of North Carolina at Chapel Hill. We gratefully acknowledge the helpful comments of Jack Vevea and Constantine Sedikides in the preparation of this article. Correspondence concerning this article should be addressed to Bernadette Gray-Little, Department of Psychology, University of North Carolina, Chapel Hill, North Carolina 27514. Electronic mail may be sent to gray [email protected]. 26

RACIAL COMPARISONS OF SELF-ESTEEM A construct from causal attribution theory, systems blame, has also been invoked to explain high self-esteem in Blacks. According to this view, Blacks are protected from low self-esteem because they are able to blame the system instead of themselves for failures and low status, whereas Whites do not have recourse to this buffer (McCarthy & Yancey, 1971). Crocker and Major (1989) also hypothesized that members of stigmatized groups can attribute negative feedback to prejudice and thereby preserve selfesteem. The systems blame interpretation seems compatible with numerous studies showing Blacks, especially children, to be lower in personal efficacy than Whites (Graham, 1994). However, systems blame is not routinely related to individual self-esteem (Gurin, Gurin, Lao, & Beattie, 1969; Hughes & Demo, 1989; Hulbary, 1975; M. C. Taylor & Walsh, 1979). Moreover, M. C. Taylor and Walsh found that although Black adults were indeed more likely than White adults to blame the system, systems blame was related negatively to self-esteem. In view of such findings, systems blame seems an unlikely explanation of a Black selfesteem advantage. Bachman and O'Malley (1984) took a somewhat different approach in attributing higher self-esteem among Black than White adolescents to an extreme responding style. As most studies reveal self-esteem scores to be negatively skewed with most respondents at or above the midpoint, extreme responding would result in higher scores, producing an artifactual selfesteem advantage. Explanations of a Black self-esteem advantage in terms of selective credulity, systems blame, and extreme responding appear to stem from the assumption that minority group members should have low self-esteem (Graham, 1994); contrary findings are then attributed to defensiveness or stylistic features. Social identity theory takes a different approach to self-esteem in minority group members (see, e.g., Brewer, 1991, 1993; Deaux, 1993; Mullen, 1991; Tajfel & Turner, 1986). From this perspective, Blacks, a numerical and status minority, are a highly distinct social group. As minority group members, individual African Americans feel separated from the majority and, at the same time, allied to others of their own group. In the quest for positive social identity, Blacks emphasize their desirable distinctiveness. Perceiving the in-group positively can fulfill the motive for positive social identity, a tendency that should increase as the group's distinctiveness becomes more prominent. The more personally germane the group's positive social identity, the more likely it is to affect individual self-esteem (Stryker & Serpe, 1994). This approach also suggests that individuals strive to increase positive distinctiveness when there has been dissatisfaction with the group's identity. Thus, Black Americans, as members of a distinctive minority group, derive a higher level of collective and eventually personal selfesteem from identification with their in-group than do members of the majority. This formulation would lead to the expectation of equal or higher self-esteem for Black Americans. In view of inconsistent findings and conflicting interpretations in this body of literature, research and discourse on the topic of racial differences and self-esteem have reached a relative impasse. Thus, for several reasons, a review of this literature seems timely and appropriate. First, we need to identify the conditions under which Black respondents exhibit equal, higher, or lower selfesteem than do White respondents. Although this strategy cannot pinpoint the causes of self-esteem, it can help to identify conditions that might influence self-esteem levels reported by Black and

27

White respondents. Second, existing summaries (e.g., Baldwin, 1979; Porter & Washington, 1979; Wylie, 1979) are dated and offer limited explanations of inconsistent research findings. Third, there never has been a quantitative review of this body of work, although a quantitative review presents the best possibility of objectively summarizing past and recent research. Previous narrative summaries have used a box-score or vote-counting strategy to determine whether studies indicated a significant directional difference. One problem with this strategy is that it ignores effect magnitude. A second is that when the true effect size is small or even moderate, vote counting becomes less able to detect the true effect as the number of studies increases (Bushman, 1994). Finally, and most important, a quantitative review facilitates analysis of moderator variables, such as participant and instrumentation characteristics, which may help explain discrepancies in previous research and aid in theory development. Conceptualization of Self-Esteem The construct of self-esteem is less inclusive than self-concept, which also encompasses nonevaluative self-descriptions and other self-processes. Although the two terms often are used interchangeably, the literature included under both terms pertains overwhelmingly to self-esteem (McGuire, 1984). Self-esteem has been described variously as a predictor of human behavior, a cue to how others react to us, and a fundamental human motive (Baumeister & Leary, 1998; Maslow, 1970). Moreover, there is substantial belief and moderate evidence that self-esteem is an index of psychological well-being (Rosenberg, Schooler, Schoenbach, & Rosenberg, 1995). Although increased attention is being devoted to the assessment of specific facets of self-esteem (see, e.g., Harter, 1985a; H. W. Marsh, 1993; Rosenberg et al., 1995), self-esteem is most often considered a general judgment of personal worth or value. The work of Cooley (1902) and Mead (1934) was seminal in generating research on race and self-esteem. Cooley was influenced by James (1890); however, their views on self-esteem differed in an important way. Whereas James linked self-esteem to a personal assessment of the ratio of success to aspirations in domains of relevance and importance, Cooley located the source of self-esteem squarely in social processes. Cooley's concept of the looking-glass self portrayed individual self-perception as a reflection of appraisal by others. Mead theorized that human beings form a picture of the generalized or referent other from experiences and, reflected therefrom, a generalized self-image. Moreover, humans develop self-knowledge or self-attitudes about each aspect of their experience. Self-esteem is the evaluative component of these attitudes. Thus, individual self-esteem, as used here, refers to the level of self-regard and reflects an appraisal process; it depends on the individual's social context and on direct experiences of either support and approbation or, conversely, negative evaluations. When a specific aspect of self-esteem is assessed--for example, self-concept of mathematics ability--evaluation in that domain is also influenced by characteristics of the comparison or reference group. For these reasons, we view the formation of self-esteem-whether linked to reflected appraisal or social comparison or measured as a general or specific construct--as explicitly anchored in social processes (Banaji & Prentice, 1994; Markus & Kitayama, 1991; Oyserman & Markus, 1993).

28

GRAY-LITTLEAND HAFDAHL

Self-Esteem and Racial~Ethnic Identity When racial comparisons of self-esteem are under discussion, individual self-esteem should be distinguished from two theoretically related constructs: ethnic or racial identity and racial esteem. Ethnic identity refers to the comprehensiveness or intensity of identification with one's ethnic group (see Phinney, 1990, 1996, for comprehensive discussions). A few studies have reported no relationship between individual self-esteem and ethnic identification in ethnic minority children and adolescents (see, e.g., Rotheram-Borus, 1990; Rowley, Sellers, Chavous, & Smith, 1998); however, a more substantial number have revealed small to moderate positive correlations (see, e.g., Parham & Helms, 1985; Phinney & Chavira, 1992; White & Burke, 1987). On balance, the two concepts, although conceptually and empirically distinguishable, are related.

Self-Esteem and Racial Esteem Racial esteem implies an evaluation of one's racial group. Racial esteem is a form of collective esteem, the perception of one's social group as valued or as comparing favorably with other social groups (W. E. Cross, 1987; Luhtanen & Crocker, 1992; Rowley et al., 1998; Tajfel & Turner, 1986; Triandis, 1989). The relationship between racial esteem and individual self-esteem has been addressed in an assortment of studies with mixed results (J. Banks, 1984; McAdoo, 1977; Paul & Fisher, 1980; Porter, 1971; Rosenberg, 1965; Spencer, 1984). In the well-known doll studies conducted by Kenneth and Mamie Clark, racial esteem was equated with racial comparisons of individual self-esteem. The doll studies, more than 30 in number and spanning a period from 1939 to 1977, typically employed a forced-choice format that required young children to choose between black and white, or dark and light, stimuli as a way of assessing racial and color preference (see reviews by Ballard & Keller, 1976; W. C. Banks, 1976; Brand, Ruiz, & Padilla, 1974; Gray-Little & Stern, 1992). Choosing a doll of one's own color was presumed to arise from own-race preference and was interpreted as positive selfevaluation. The most unambiguous finding from this body of research was that White children showed more own-race preference than did Black children, which was interpreted as indicating a White self-esteem advantage. A second reliable finding was that racial differences dissipated among older participants due to agerelated increases in own-race preference among Black children. Racial esteem as measured by the doll test failed to predict children's actual choice of playmates and friends (Finkelstein & Haskins, 1983; Hraba & Grant, 1970; Rosenberg & Simmons, 1972; Schofield & Sagar, 1977; Silverman & Shaw, 1973; Singleton & Asher, 1979) and was inconsistently related to more direct measures of self-esteem (Clark, 1982; W. E. Cross, 1987; Hines & Berg-Cross, 1981; Houston, 1984; Rosenberg, 1981; Ward & Braun, 1972). Interest in the doll test has subsided, and only a handful of recent studies has been devoted to this topic (see, e.g., GoPaul-McNicol, 1988; May, 1981; McMillan, 1988; PowellHopson & Hopson, 1988). Most important, the doll studies did not directly assess racial differences in self-evaluation, the major focus of the present review.

Racial Differences in Self-Esteem First, we address the general question "Are there consistent racial differences in self-esteem?" Our review of the empirical and theoretical literature led us to predict higher average self-esteem among Black respondents. However, in view of substantial inconsistency across studies, we also attempted to identify moderators associated with discrepant findings. Thus, we posed the following more specific questions: Do racial differences vary as a function of participant characteristics such as age, gender, and social class, or of the racial make-up of the settings from which participants are drawn? Are such features of self-esteem instruments as quality, domain of content, or response format related to racial comparisons? Are aspects of the research report such as the year or type of publication associated with the direction or magnitude of the difference found? Finally, how are these moderators interrelated? Overviews of the moderators and associated predictions are presented in the following sections.

Characteristics of the Participants and Their Social Context Participant age. Predictions regarding age should stem from knowledge of both developmental patterns in self-esteem and racial differences in those patterns. However, research on both topics is equivocal. Wylie (1979) concluded that there were no consistent age differences in self-esteem. Several studies of children and adolescents have been published since Wylie's review or were not included in her review. Eight of 11 longitudinal studies showed an increase in self-esteem with age (Barnes & Farrier, 1985; Fu, Hinkle, & Korslund, 1983; Hare, 1985; Jessor & Jessor, 1977; McCarthy & Hoge, 1982; Mullis, Mullis, & Normadin, 1992; O'Malley & Bachman, 1983; Wolf, Hunter, Webber, & Berenson, 1981). The remaining longitudinal studies--1 with a primarily White sample (Hoge, Smit, & Hanson, 1990) and 2 presenting data separately for racial groups (Gray-Little & Appelbaum, 1979; Wade, Thompson, Tashakkori, & Valente, 1 9 8 9 ) found no evidence of increased self-esteem with age. Cross-sectional studies have revealed age-related increases in self-esteem in a predominantly Black sample of fifth through eighth graders (Demo & Savin-Williams, 1983) and increased self-esteem among Black, but not White, adolescents (Dales & Keller, 1972). Twenge's (1997) analysis of 172 studies conducted with the Rosenberg Self-Esteem Scale (RSE) revealed increases in self-esteem with age, with the exception of a dip in self-esteem levels during junior high school. The pattern of increasing selfesteem with age was stronger for male than for female participants but did not differ significantly by racial group. Her analysis of the 139 studies using the Self-Esteem Inventory (SEI) showed a roughly similar pattern. Harter (1982) also reported a decline in self-esteem in seventh-grade children. Cross-sectional analyses by Gray-Little and Appelbaum (1979) and Mullis et al. (1992) revealed no age-related increases in self-esteem. In summary, crosssectional and longitudinal research suggests increases in selfesteem in adolescence, possibly following a decline in preadolescence, and perhaps a greater increase for Blacks than for Whites. Thus, we expected a positive association between age and effect size (ES; standardized Black-minus-Whitemean difference). Sex. In their influential review of research on sex differences, Maccoby and Jacklin (1974) concluded that there were no sex

RACIAL COMPARISONS OF SELF-ESTEEM differences in self-esteem. Wylie (1979) also judged thenexisting research inadequate to draw conclusions about sex differences in self-esteem. However, Skaalvik's (1986) review of 29 studies published between 1975 and 1985 revealed strong evidence that among children and adolescents, boys receive higher self-esteem scores than girls on a variety of standardized and unstandardized instruments. Later investigations by Josephs et al. (1992), McRae (1991), and O'Brien (1991) also indicated higher general self-esteem for male participants. Several investigations of the joint associations of race and sex with self-esteem suggest that the male self-esteem advantage is more common among Whites than Blacks (Lawrence & Brown, 1976; Martinez & Dukes, 1987; Richman, Clark, & Brown, 1985; Simmons & Blyth, 1987; Turner & Turner, 1982). For that reason, we expected that the percentage of female participants in the sample would be associated with a greater Black selfesteem advantage, resulting in a higher positive ES. Socioeconomic status. Socioeconomic status (SES) is often positively associated with self-esteem, but less for Blacks than for Whites (Demo & Savin-Williams, 1983; Rosenberg, 1981; Rosenberg & Pearlin, 1978). Explanations for the weaker relationship between SES and self-esteem among Blacks focus on the greater viability of protective strategies--insulationand rejection of mainstream values--for this group. These mechanisms are presumed to be more viable for lower- and working-class Blacks than for those in the middle and upper classes because of better insulation from unfavorable social comparison (Heiss & Owens, 1972; McCarthy & Yancey, 1971). Although this explanation can be debated, empirical patterns found in previous research suggest a greater self-esteem advantage for Blacks among low than among middle or high SES groups. Racial consonance. Social context may be consonant or dissonant in many ways (e.g., race, gender, religion, or SES). These features of the social context may influence perceived appraisals, social comparisons, and judgments of competence and may be associated with the size and direction of group differences in self-esteem. For example, Rosenberg (1965) reported that ethnic consonance enhanced self-esteem: Catholic, Jewish, and Protestant youth from neighborhoods where their respective ethnic groups constituted the majority had higher self-esteem than did corresponding adolescents who constituted a minority group in their neighborhoods. In a national study, J. S. Coleman (1966) found that Blacks in segregated schools had higher self-esteem than those in integrated schools, a conclusion that was consistent with the findings of St. John (1975) and Stephan and Rosenfield (1978). Although not all investigators have reported an association between self-esteem and racial consonance (see, e.g., Epps, 1975; Jensen, White, & Galliher, 1982), additional support for this association comes from studies published in the 1980s and 1990s (Abbot, 1981; Gerken, Allen, & Snider, 1984; Gray-Little & Carels, 1997; Powell, 1985). Thus, the preponderance of evidence, though not entirely consistent, supports the expectation of a positive association between racial consonance and self-esteem differences--a positive ES in settings of racial consonance for Black participants and a smaller or negative ES in settings of racial consonance for White participants.

29

Instrumentation Standardization. Our goal in examining standardization was to conduct a rudimentary study of the association of instrument quality with racial differences in self-esteem. Instrumentation is critically important in this area of research because the central or only manipulation in many relevant studies was respondent race. Hence, instrument quality and adherence to standard administration and scoring procedures were the primary determinants of data quality. There is little research directly investigating the relationship between instrumentationand racial differences in self-esteem. However, Wylie (1979) concluded that "studies reporting lower black self-esteem tend to be methodologically less adequate than those reporting them to have higher self-esteem" (p. 182). Standardization is only one aspect of methodological adequacy; however, it establishes a minimal baseline for quality. (See Demo, 1985, and Wylie, 1989, for reviews of self-esteem instruments.) On the basis of previous reviews suggesting a Black self-esteem advantage and of Wylie's conclusion, we expected racial comparisons derived from standardized instruments to yield larger ESs than would those from unstandardized instruments.l Comparison of four standardized instruments. In view of the subjective nature of self-esteem, it is not surprising that self-report is the typical measurement approach. In adolescents and young adults, self-esteem is measured most often by combining true-false or Likert-type responses to questionnaire items into a total score. Among younger children, similar measures, sometimes administered orally, along with interviews and pictorial formats that allow nonverbal responses are common (Harter, 1982, 1985b; Piers, 1984; Wiley, 1989). Standardized questionnaires were used frequently in the studies we identified; however, the equivalence of even standardized measures has been debated widely (Demo, 1985; Wiley, 1989). Thus, it seemed appropriate to evaluate the comparability of standardized instruments that have been used often in assessing racial differences in self-esteem. The four instruments--RSE, the Coopersmith SEI, the Tennessee SelfConcept Scale (TSCS), and the Piers-Harris Children's SelfConcept Scale (PH)--together account for approximately 60% of the measurement-related citations in the literature on self-esteem from the mid-1960s to 1990 (Blascovich & Tomaka, 1991). There are important conceptual similarities, but also conceptual and structural dissimilarities, among the measures. Each purports to assess general or overall self-esteem rather than a specific facet of self-esteem. Whereas the RSE (Rosenberg, 1965) assesses general self-esteem without reference to specific domains of functioning such as family or social life and can thus be described as a global instrument, the other three assess general self-esteem by summing across evaluations of one's assets and liabilities in various domains and can thus be described as aggregate instruments. The instruments also vary in length, target age-range, and response format. The RSE, a 10-item, global, Likert-type instrument that is typically used with respondents from junior high school age through adulthood, has been the focus of intense psychometric One criterionused to selectindependentcomparisonsfrom studieswith multiple comparisons was the choice of standardized instruments over unstandardized ones. However, in no instance was a selection made primarily with this criterion.

30

GRAY-LITTLEAND HAFDAHL

evaluation and offers a highly reliable and internally consistent measure (Gray-Little, Williams, & Hancock, 1997). Despite being negatively skewed, the RSE is reliably related to measures of overall well-being (Rosenberg et al., 1995). The SEI is a 50-item, aggregate, true-false instrument that yields a total score and subscores in the areas of peers, parents, school, and personal interests (Coopersmith, 1967). A primary problem with the SEI is that the factor structure has not been stable across several studies of the instrument; nor do factors identified in various studies of the instrument always correspond to the four conceptual areas the instrument was developed to address. Correlations between the SEI and RSE range from .58 to .60 (Blascovich & Tomaka, 1991). The PH is an 80-item, aggregate, true-false instrument that assesses total self-esteem and six factor analytically derived clusters: school status, physical appearance and attributes, anxiety, popularity, happiness, and satisfaction (Piers, 1984). Correlations between its total score and that of the SEI range from .63 to .85 (Wylie, 1989). Although the PH was designed for use with children and adolescents, it seems better suited for younger children because of increased social desirability when the scale is used with adolescents. The TSCS is a 100-item, aggregate, Likert-type scale designed to provide a total positive score as well as an indication of self-esteem in five areas: moral-ethical, personal, physical, family, and social self (Fitts, 1965). Factorial studies have indicated validational support for the family, social, and physical scales (H. W. Marsh & Richards, 1988). The TSCS was intended for use with respondents 12 years of age and older. A correlation of .75 has been reported between the total scores of the TSCS and SEI (Blascovich & Tomaka, 1991). There are published data on reliability and construct validity for each instrument (Blascovich & Tomaka, 1991; Wylie, 1989). Each has acceptable levels of internal consistency and test-retest reliability, and each has been criticized as eliciting social desirability. Among the four instruments and in comparison with most other measures of general self-esteem, the RSE is often considered the gold standard. Because of its desirable psychometric features (and the expectation that stronger instruments reveal a self-esteem advantage for Black respondents), we expected a relatively larger ES size from studies using the RSE. However, the RSE also differs in being a global instrument, and finding a larger ES with the RSE might be attributable to this feature. Thus, an important question for this comparison was whether the self-esteem advantage for Black respondents is limited to the RSE. Response format. Bachman and O'Malley (1984) argued that the Black self-esteem advantage was artifactual, attributable to an extreme responding style that can be expressed on Likert-type scales. They attempted to obviate the effects of extreme responding by truncating all positive response options into one category and all negative options into another. With this procedure, racial differences in self-esteem were eliminated. However, in a negatively skewed distribution, collapsing all positive scores into one category may well obscure valid differences between groups varying in levels of positive self-esteem. Bachman and O'Malley's study was based on a review of studies using various forms of the RSE, a global instrument, with high school students. To examine the extreme responding hypothesis, we contrasted results from the two aggregate instruments (SEI and PH), whose true-false response format precludes extreme responding, with those from the TSCS, which is also an aggregate scale but with a Likert-type

response format. Indirect support for Bachman and O'Malley's hypothesis would be demonstrated by a larger positive ES on the TSCS, which is liable to extreme responding, than on scales where responses are constrained to two choices. Domain. The magnitude and direction of racial differences in self-esteem can vary with the specific area of self-esteem being assessed (Tashakkori, 1993). Thus, the content or domain assessed by an instrument may be important for understanding discrepancies in the literature on race and self-esteem. Most studies reviewed here used either domain-free global scales or aggregate scales, both of which assess general self-esteem. Others, however, employed scales designed to measure self-esteem in specific domains, for example, academic or physical self-esteem. Although the use of general self-esteem instruments aligns better with the traditional notion of self-esteem as overall self-regard, domainspecific instruments offer some advantages. Reflecting the basic principle that the power of an attitude to predict behavior is a function of how closely the attitude relates to the behavior (Fishbein & Azjen, 1975), specific instruments are better able to predict competence in the targeted domain. Thus, persons of high musical or athletic ability are likely to have higher self-esteem in those areas than individuals with low ability in those areas (Felson, 1993; Rosenberg et al., 1995), but they may not differ in general self-esteem. Further, as specific self-esteem is more related to competence, it also may better predict self-efficacy in particular areas (Harter, 1985a; Hoge et al., 1990; H. W. Marsh, 1986, 1993). We examined racial differences only in academic versus general self-esteem because other specific domains were reported infrequently.2 Relatively higher self-esteem scores for Black respondents were expected with the use of general instruments (global and aggregate) than with specific, academic self-esteem instruments. This expectation was based on the following reasoning: Numerous studies show higher academic achievement for White than for Black students (see, e.g., J. S. Coleman, 1966; Steele, 1992). Because academic achievement and academic self-esteem are positively related (Hansford & Hattie, 1982; Skaalvik & Hagtvet, 1990), we would expect higher academic self-esteem for White students. Although some contradictory results have been reported (see, e.g., Hare, 1980; Kugle, Clements, & Powell, 1983), several investigators have found support for this expectation (see, e.g., Lay & Wakstein, 1985; Mboya, 1988; W. L. Osborne & LeGette, 1982; Richman et al., 1985).

Characteristics of the Authors and the Research Report Author race. Experimenter or author race and ethnicity have been related to outcome in research ranging from studies of social attitudes to physiological reactivity (Anderson, 1989; Gray-Little, 1980; Rosenthal, 1994; Sattler, 1970). First, experimenter race may affect participants' reactions as demonstrated in the doll studies and other research (e.g., Gray-Little & Teddlie, 1978; Murphy, Alpert, Moes, & Somes, 1986). Second, research investigators themselves are not immune to biases that may affect the 2 Many aggregate scales include school-related items, but such items typically constitute a small portion of the scale content. Studies using aggregate scales commonly reported only the total score, which did not allow retrieval of subscale scores. Thus, our comparison is limited to general versus academic self-esteem.

RACIAL COMPARISONS OF SELF-ESTEEM conduct and outcome of research (Greenwald & Schuh, 1994). Biases relevant to the present research include expectancy effects that can be communicated to research participants and belief perseverance in the face of discredited information, which might affect interpretation of results (Miller & Pollock, 1994). Third, the race of the author may influence outcome in other inadvertent ways, for example, practical considerations such as the sample chosen or the type of samples to which investigators gain access may be affected by their race or reactions to their race. We anticipated that the ES would be positive when authors were Black and smaller or negative when authors were White. Year of publication. What is the relationship between the time of data collection and racial differences in self-esteem? Social changes may not only influence individual levels of self-esteem but also affect groups differently (Caspi, 1987; Stewart & Healy, 1989; Tajfel & Turner, 1986; Twenge, 1997). Self-esteem is a prime target for such change because often it is viewed as central to health and happiness, and self-esteem enhancement is endorsed heartily by the popular culture (Swarm, 1996). More to the point, many programs aimed at improving the social, academic, and economic status of African American children, adolescents, and adults have targeted increases in self-esteem as a key ingredient to both individual and group advancement. Initiatives such as the "Black is beautiful" movement were anchored in the belief that changing the value associated with characteristics of Blacks might enhance both individual and group esteem among Blacks. Although the efficacy of this type of program in raising self-esteem has not been demonstrated explicitly, the Civil Rights Movement of the 1960s and 1970s has often been credited with enhancing self-esteem among Black children and adults (Baumeister, Smart, & Boden, 1996; Powell & Fuller, 1973). Following this logic, it would be reasonable to predict an enhancement of Black selfesteem relative to Whites over the period covered by this review; that is, publication date and ES should be positively related. Publication outlet. The purpose of this analysis was to examine the association between the source o f data and ES. This classical methodological consideration in meta-analysis is especially germane to the present study, where almost equal numbers of ESs were from two relatively nonoverlapping sources: Most dissertations are not published, and only a small proportion of published studies result from dissertations. To the extent that the sources represent two different bodies of data, ESs found in the two sources might differ in magnitude. Method

Eligibility Criteria The meta-analytic sample comprised research published between 1960 and June 1998 on racial differences in self-esteem among children, adolescents, and college-age adults. The numerous studies of self-esteem conducted with these age groups ensure good representation of these populations. We accessed reports through the following sources: (a) citations from published empirical studies and narrative reviews on Black self-esteem and on race and self-esteem; (b) computerized searches of the PsyclNFO, Sociological Abstracts, Medline, and ERIC databases using the terms self-esteem, self-concept, or self with race, and race names; and (c) manual searches of relevant journals. Studies were eliminated from further consideration if the information reported was not adequate for the calculation of a standardized ES, clinical or incarcerated populations were used, or the average age of the sample exceeded 22 years.

31

Only independent comparisons were included in the present analysis: More than one comparison from a single study was used only when the different comparisons included different samples. Our selection of the independent comparison for inclusion was guided by the following four criteria, listed in roughly descending priority: We chose (a) instruments developed specifically for measuring self-esteem over those developed for other purposes (e.g., a scale of adjustment), (b) objective instruments over those that required interpretation before scoring, (c) standardized instruments with published reliability data over unstandardized instruments, and (d) the first testing over subsequent testings, when participants were tested more than once. The choice between two objective, standardized instruments of self-esteem administered at the same time was made at random but was rarely necessary.

Data Collection The following information, when available, was recorded for each comparison: (a) all results needed to compute an ES (the standardized mean difference between the self-esteem scores of Black and White participants); (b) the numbers of Black and White participants; (c) the percentage of each race in the school(s) from which each subsample was drawn; (d) the numbers of male and female participants in the sample; (e) participants' age (when grade level was reported rather than age, the former was converted to an age equivalent by adding five to the grade level); (f) the name, domain (general or specific type), standardization status, and format (e.g., questionnaire, checklist) of the self-esteem instrument; (g) the years of data collection and of publication; (h) SES; (i) the first author's race; and (j) publication outlet, whether a journal or book versus a dissertation. The preceding items were recorded separately for each subsample when the original study reported information separately for each subgroup (e.g., sex, age, SES). For published studies, we requested missing information from the author(s).

Data Coding For most variables (e.g., age, number of male and female participants, grade level), data were presented numerically in the original publication and did not require coding; accuracy was monitored by having a second person check data entries. Different procedures were used for SES and instrument standardization. SES information often was missing from the studies or was reported imprecisely. Two independent coders who were not familiar with the hypotheses of the study were trained to code SES as high, middle, or low. SES was coded most frequently from authors' description of the sample as high, middle, or low SES. In the absence of the authors' categorization, coders used information such as the Hollingshead and Redlich (1957) or Duncan (196l) indexes or the characterization of the school or neighborhood from which the sample was drawn as being of a particular social class. Cohen's kappas for the 129 rated ESs were .93 (unweighted) and .94 (weighted, with disagreements by one or two SES levels weighted as 1 and 2, respectively). The following rules were used to classify instruments as standardized or unstandardized. Instruments were classified as standardized (k = 177 ESs) when they had an established set of items and a uniform procedure for administration and scoring (Anastasi & Urbina, 1997). In addition, most standardized tests had norms for one or more age groups and published information on such psychometric characteristics as reliability and validity. Typically, instruments classified as standardized were well-known instruments, for example, the RSE, TSCS, or SEI. Instruments developed for use in a particular study, with little or no data on psychometric qualities, or those comprising idiosyncratic combinations of items from standardized scales, often together with the author's items, were classified as unstandardized (k = 84 ESs).

Statistical Methods ES computation. The ES index was the standardized mean difference estimate d--the unbiased estimator based on Hedges's g (Hedges & Olkin,

32

GRAY-LITTLE AND HAFDAHL

1985), calculated such that positive values indicate a Black self-esteem advantage and negative values, a White self-esteem advantage. The correction for bias uses the total sample size and g. Depending on the information reported, we used one of the approaches described below, in descending order of preference, to compute g: 1. Most gs (224 of 261 independent ESs) were computed from Black and White participants' subsample sizes and self-esteem means and standard deviations. 2. For 10 ESs where group means but no standard deviations were reported and an MSe~,or could be extracted from an analysis of variance (ANOVA) with race as a factor, we used the MSerror to compute the pooled standard deviation for the denominator of g. When factorial ANOVA results were reported in sufficient detail, we adjusted the reported MSer~o~ by returning other between-groups sources of variability to the error term. 3. When the t statistic (11 ESs, including point-biserial correlations) or F statistic (12 ESs) was reported for a Black-White self-esteem comparison, we converted the test statistic to g. 4. When no test statistic was given but a comparison was reported as significant (4 ESs), we determined the critical t value corresponding to the reported alpha level (assuming two-tailed a = .05 unless reported otherwise) and followed the procedure in 3 above. 3 5. When a comparison was reported only as nonsignificant (16 ESs) and we were unable to compute an ES, we excluded the comparison from further consideration. A sensitivity analysis conducted to examine the consequences of this decision showed that it did not appreciably alter results or substantive conclusions.4 ES weights. For most analyses, each ES was weighted. Initially each ES was weighted by the inverse of its conditional sampling variance (Hedges & Olkin, 1985; Shadish & Haddock, 1994). Such weights are proportional to sample size, reflect the precision of each ES as an estimate of the population ES, and are optimal in that they provide a minimum variance estimate. However, optimal weighting was problematic given the current data set, which included seven studies (Collins, 1993; Griffith, 1985; Keith, Pottebaum, & Eberhart, 1986; Kohr, Coldiron, Skiffington, Masters, & Blust, 1988; Lay & Wakstein, 1985; J. W. Osborne, 1995; Tashakkori & Thompson, 1991 ) whose 15 comparisons involved extremely large samples. These 15 ESs amounted to less than 6% of the 261 ESs but contributed 86% (542,129) of the 632,872 total participants in the sample and 80% of the total optimal weight. Under such circumstances, a few heavily weighted studies--which may not be representative of either the identified or unidentified characteristics of the complete literature--can dominate the analysis. To curb the influence of these few studies, we imposed an upper limit of 700 on ES weights; on the basis of visual inspection of the weight distribution, this value is slightly greater than optimal weights for the 246 ESs from smaller samples (which ranged from 2.19 to 613.41, with M = 52.84, Mdn = 27.79, and SD = 85.78). With the restricted weighting, the same 15 ESs accounted for only 45% of the total weight. We conducted a sensitivity analysis to examine the impact of this weight restriction and found that most results and conclusions would have been essentially the same using optimal weighting for all ESs. 5 Analyses of ESs. After examining the weighted and unweighted ES distributions, we combined the weighted ESs to estimate the population mean ES and assess homogeneity. When ESs were not homogeneous, we examined their heterogeneity by relating them to study characteristics. For each characteristic, or moderator, we conducted an analysis using only the subset of cases with complete information for that moderator. Analyses were carried out with SPSS for Windows (SPSS, 1997), Microsoft Excel (Microsoft, 1997), and a hand-held calculator, following procedures described in Hedges (1994) and Hedges and Becker (1986). Continuous moderators. Most continuous moderators (e.g., participants' age, proportion of male participants, racial make-up of the sample) were examined with a simple linear regression analysis, with each ES weighted as described above. The analysis of school racial consonance, however, involved two moderators--proportion of Black students in the

school(s) from which Black participants were drawn and proportion of White students in the school(s) from which White participants were drawn--in a multiple regression. The results of both the simple and multiple regression analyses include point and interval estimates of the population regression coefficient for each moderator and a heterogeneity statistic to assess the model's adequacy. 6

3 This procedure may introduce a downward bias in the results. However, as only 4 ESs were calculated in this manner (and the sample sizes--170, 186, 212, and 249--were relatively small), it would not have a large biasing effect on the overall analysis. 4 To assess the impact of these exclusions, we substituted an ES of d = 0.00 for the 16 excluded comparisons, which, if included in our sample, would have contributed only 6% of the comparisons, 1% of the total participants, and 6% of the (restricted) weights. The overall distribution of unweighted ESs would have had M = .110, Mdn = .110, SD = .359, Qt = -.118, and Q3 = .349. After weighting, the overall mean ES would have been estimated as do = . 140, and the weighted ES distribution would have had Mdn = .201, SD = .262, QI = - .041, and Q3 = .315. For both categorical and continuous moderators, the estimate magnitudes moved slightly away from zero; however, similar significance levels obtained for each estimate. For categorical moderators, the direction of all total sample and subgroup mean ES estimates remained the same when these nonsignificant comparisons were excluded. Further, the significance of all heterogeneity tests remained essentially the same as before the exclusion. For continuous moderators, the direction of all regression coefficient estimates remained the same with the exclusion. Finally, although each Q~sidual decreased slightly, it remained vastly significant for every moderator. 5 Without these weight restrictions, the overall mean ES would have been estimated as a slightly lower do = .130. For categorical moderators, the direction of all total sample and subgroup mean ES estimates remained the same with the weight restriction. Although most ES magnitudes changed slightly with the restrictions, a few more than doubled: the overall mean ES for the author race and SES analyses and subgroup mean ESs for White authors and for low SES samples. The significance levels of all mean ES estimates were virtually unchanged, and the rank order of moderator subgroup mean ESs remained the same. The Qwithin and all within-group Q statistics necessarily decreased when weights were restricted, but the significance of most heterogeneity tests, including tests of Qbet. . . . . remained essentially the same as before the restriction. Exceptions to this pattern obtained for author race and participant sex. Without the restriction, White authors had a smaller d. = .038, yielding a trend toward a significant author race difference, abetween = 3.75, p = .053; and female participants had a larger do = .254, resulting in a significant sex difference, Qbetween = 22.42, p < .001. For all continuous moderators except publication year, the direction of regression coefficient estimates remained the same with the restriction. The slope for publication year before restriction was a significantly negative ~ = -.002, with a 95% confidence interval of (-.004, -.001); this became nonsignificantly positive with the restriction. Although all slope estimates' magnitudes moved toward zero with the restriction, each estimate's significance remained essentially the same (except for publication year). Finally, Q~siaual necessarily decreased but remained significant for every moderator. 6 In the context of meta-analysis, a weighted regression analysis uses within-study sample size information from each individual study. This allows a comparison of the observed to the expected heterogeneity among ESs, given the sample sizes and variances. This test of the model's statistical adequacy, which addresses a different issue than the more familiar coefficient of determination (R2), cannot be conducted in the ordinary least squares regression analyses commonly used in the analysis of primary data.

RACIAL COMPARISONS OF SELF-ESTEEM Categorical moderators. We examined each categorical moderator (participant sex and SES; instrument standardization, domain, and response format; publication type; and author race) separately by conducting a fixed-effects ANOVA-type analysis. This analysis produces estimates of the overall and group mean ESs, as well as several overall and group heterogeneity statistics, to assess the model's statistical adequacy. The finding that a group mean ES varied significantly (i.e., there was significant between-group heterogeneity) was followed by specific group comparisons to identify the source(s) of between-groups variability. ES differences between specific moderator groups were examined by planned contrasts as described by Hedges (1994). Study-level analyses. Although practical obstacles, primarily missing moderator data, prevented a thorough examination of factorial categorical designs and multiple regression models, we did examine associations between certain moderators using contingency-table analyses and chisquare tests of independence for categorical moderators, ANOVAs for categorical and continuous combinations, and correlations for continuous moderators. For these analyses, which addressed study- rather than participant-level characteristics and did not include ESs, each comparison was weighted proportionally to its restricted ES weight to reflect its relative influence in the primary moderator analyses.

Results Primary Studies Analyzed and Those Omitted Published studies. A total of 261 independent ESs were available from both published studies and dissertations. Here we refer to the 135 ESs obtained from studies published in journals (98%) and in books (2%). Of the 58 authors to whom we initially sent letters requesting information missing from the study reports, 34 responded. Follow-up letters sent to 15 authors led to 5 additional replies. Telephone and E-mail requests to 10 additional authors resulted in 7 more replies. Three studies were eliminated because they did not include information necessary to calculate an ES; we also excluded 1 study because the age of the sample was over 22 and 2 others because the respondents were incarcerated or clinical populations. A total of 135 independent ESs based on comparisons from 72 studies were used in the meta-analysis; 29 (40%) studies contributed more than one independent ES. Dissertation research. We identified 63 dissertations that appeared to contain information on racial comparisons of selfesteem. We obtained copies of 46 dissertations through intedibrary lending, 11 were purchased from University Microfdms, and 6 were reviewed in visits to the relevant libraries. Forty-eight of the dissertations were found to have usable data, 5 presented no racial comparisons, 8 did not include information necessary to calculate an ES, and 2 were excluded because of the age of the sample. A total of 126 independent ESs from the 48 dissertations were used in the meta-analysis; 21 studies (44%) contxibuted more than one independent ES. O v e r a l l R a c i a l D i f f e r e n c e in S e l f - E s t e e m

The unweighted ES distribution was fairly symmetric and unimodal, with M = . 117, Mdn = . 155, SD = .369, Q1 = - . 135, and Q3 = .362. The 261 ESs included 632,872 participants; the number of participants per ES was strongly positively skewed, with M = 2,424, Mdn = 137, Q1 = 52, and Q3 = 338. The overall mean ES, do = .148, is significantly positive and indicates slightly higher self-esteem for Black than White partici-

33

pants; on the basis of SE (2°) = .007, the 95% confidence interval (CI) about the population mean was (.135, .161). A CI that excludes zero provides evidence of statistical significance. The obtained effect size is relatively small but indicates that, on the average, the selfesteem score of Blacks exceeds that of Whites by approximately .15 standard deviation units. We estimated this overall population standardized mean difference using the distribution of weighted ESs, which was somewhat negatively skewed with Mdn = .208, SD = .267, Q1 = -.076, and Q3 = .315. The heterogeneity statistic for the 261 ESs is Q (260) = 1,673.72, which is significant when referred to a chi-square distribution, p < .001, and indicates significant unexplained heterogeneity among the ESs. We attempted to understand this variability by relating it to moderators that seemed important for theoretical or methodological reasons. Moderator Analyses Continuous moderators. Table 1 contains distributional information for all continuous moderators. (See the Appendix for a more complete listing of continuous and categorical moderator data.) For each moderator, we report the number of ESs, the estimated population regression coefficient (i.e., the predicted ES increase for a unit increase in the moderator), the standard error of this estimate (the usual weighted least squares standard error divided by the residual mean square; see Hedges, 1994), and a 95% CI about the corresponding population regression coefficient. The residual homogeneity statistic, aresidual (with d f = k - p - 1, where p is the number of moderators), indicates whether heterogeneity remains unexplained by the linear model. Categorical moderators. Table 2 contains frequency information for all categorical moderators. For each categorical moderator, we report the following results for the total group and for subgroups: the number of ESs and the estimated population mean ES, the standard error of this estimate, and a 95% CI about the population ES. We also report the heterogeneity test statistic Q (distributed as chi-square on d f = k - 1) with its significance level. A significant Q indicates heterogeneity among the ESs beyond that expected given their conditional variances. A significant Qb,t . . . . for the between-groups source (with d f = p - 1, where p is the number of moderator groups) indicates heterogeneity among mean effects (i.e., moderator groups differ from one another). Qwithin for the within-groups source (with d f = k - p, where k is the total number of ESs in the subset analyzed) provides a test of pooled within-groups heterogeneity and, therefore, of the model's adequacy. A significant Qwithin indicates that additional heterogeneity remains unexplained by the moderator. The test statistic for all planned contrasts of mean effect differences between specific moderator groups was referred to the chi-square distribution with d f = 1. C h a r a c t e r i s t i c s o f the P a r t i c i p a n t s a n d T h e i r Social Context Participant age. The analysis of age effects is presented in Table 3 and confirms our expectation of a positive association between ES and participant age. This is a substantial effect and indicates an ES increase of .030 for every year of increasing age. This association is illustrated in Figure 1. In studies with the youngest participants, the ESs indicate higher White self-esteem;

34

GRAY-LITTLE AND HAFDAHL Table 1 Summary Distribution Information for Continuous Moderators School(s) proportionb

Sample proportion" Statistic

Participant age

k Maximum 03 Mode 2 (k) Mode 1 (k) Mdn M Ql Minimum SD

257 21.00 15.50 16.00 (20) 13.00 (20) 13.00 12.46 10.00 3.50 3.80

P(male) 235 1.00 1.00 1.00 (77) 0.50 0.52 0.00 0.00 0.39

P(Black) 261 0.79 0.50 0.50 (59) 0.47 0.40 0.27 0.03 0.18

P(Black)

P(White)

90 1.00 0.49

89 1.00 0.86 1.00 (12) 0.16 (12) 0.73 0.65 0.53 0.16 0.27

0.27 (14) 0.27 0.39 0.20 0.04 0.30

Publication year 261 1997 1985.50 1977 (38) 1979.00 1979.51 1973.00 1965 6.99

Note. k = number of independent comparisons. Modal frequency follows mode in parentheses. P(.) = proportion of • in the sample or school(s). Proportion base rate is the sample of respondents included in the comparison, b Proportion base rate is the entire school(s) from which the respective racial subsample was drawn; for some studies, Black and White students were drawn from different schools.

Table 2 Frequency o f Categorical Characteristics Characteristic/group Participant sex Female Male Mixe~ Missing b Socioeconomic status Low Middle High Mixed~ Missingb Standardized instrument Coopersmith Piers-Harris Rosenberg Tennessee Semantic differential SOS Self-acceptance c Other Instrument standardization Standardized Unstandardized Instrument domain General Academic Other Author race Black White Hispanic Other Missing b Publication outlet Journal/book Dissertation

k 63 77 95 26 76 42 11 20 112 37 26 18 30 4 12 50 177 84 217 37 7 19 81 8 9 144 135 126

Note. k = number of independent comparisons. Italicized groups were excluded from analyses. The sample included participants of both sexes or from two or more socioeconomic statuses, b Information was not provided in the report and could not be obtained from the author(s), c Self-acceptance scale from the Self Observation Scales (Gregory, 1997).

however, the direction appears to reverse by age 10. Thereafter, there is a self-esteem advantage for Black respondents. Sex. We hypothesized a positive association between ES and proportion of female participants in the sample. The relationship of sex to racial differences in self-esteem was examined in two ways. First, the 140 ESs that included only male or only female respondents were compared. The average ES from both sexes was positive. Although the ES obtained with female participants was larger than that from all-male samples, this difference was nonsignificant (see Table 4). In an alternative analysis that included a larger number of comparisons (k = 235), ES was regressed on the proportion of male participants in the sample. The proportion of male participants in the sample was negatively related to ES (see Table 3); as the proportion of male participants increased from 0.0 to 1.0, the ES would be expected to decrease by .042, but this association was also nonsignificant. Thus, both analyses were consistent with, but did not confirm, the prediction that the selfesteem advantage for Black respondents would increase in studies with a higher percentage of female participants. Age and sex. To further explore age and sex effects, we assessed the association of ES with the interaction of age and sex (see Table 3 and Figure 2). The significant negative interaction between proportion of male participants in the sample and age indicates that, at younger ages, there is a greater self-esteem advantage for Whites when the sample is all female; in early adolescence, ES seems less related to participants' sex; thereafter, however, the Black self-esteem advantage increases more rapidly in samples with more female participants. Thus, the predicted increase in Black self-esteem advantage in increasingly female samples appears to occur only after preadolescence. SES. Because of the importance of SES in understanding the social context of participants, we have analyzed its association to ES. However, SES data were often missing or presented in an ambiguous way. Thus, we urge extreme caution in interpreting the results, which, we believe, may understate the role of SES. We expected a larger positive ES in low than middle or high income groups. The ES is largest for the low SES sample (see Table 4), with a smaller but still positive ES for the middle SES group. For

RACIAL COMPARISONS OF SELF-ESTEEM

35

Table 3

Continuous Moderator Analyses: Participant Characteristics and Study Context Moderator

~

SE(~)

Participant age P(male) a Age and P(male)b Participant age P(male)a Age X P(male) P(Black) a Racial consonance P(Black)c P(White)c Publication year

.030 -.042

.002 .025

(.026, (-.091,

.030 -.036 -.018 .157

.002 .026 .008 .034

(.026, .034) (-.088, .015) (-.034, -.003) (.090, .224)

.310 - .302 .001

.053 .061 .001

(.207, .413) (-.422, - . 182) (-.001, .003)

Qresid

df

1,476.93"* 1,313.26"* 1,132.47"*

255 233 227

1,653.26"* 367.20**

259 86

1,672.34"*

259

95% CI for/3 .034) .008)

Note. ~ = weighted least squares unstandardized regression coefficient estimate; SE(~) = standard error of ~; 95% CI for /3 = 95% confidence interval about the population regression coefficient; Qresid = residual heterogeneity statistic (distributed approximately as chi-square), representing between-effect sizes heterogeneity not explained by the regression; df = degrees of freedom for Qr¢,id; P(') = proportion of • in the sample or school(s). Proportion base rate is the sample of respondents included in the comparison, b Both predictors were centered at zero before constructing the interaction tenn. c Proportion base rate is the entire school(s) from which the respective racial subsample was drawn; for some studies, Black and White students were drawn from different schools. • * p < .0001.

the high SES sample, the ES is negative, indicating a self-esteem advantage for White respondents. Planned contrasts indicated that the low SES group differed significantly from the combined middle and high groups, g = .201, X2(1) = 19.62,p < .001; and that the latter two groups differed from one another, g = .126, )(2(1) = 33.31,p < .00l. 7 Thus, the results confirm the expectation of an inverse relationship between ES and SES. Racial consonance. Racial consonance was examined first in terms of the racial make-up of the sample (proportion of Blacks) and then with regard to the racial make-up of the setting from which participants were drawn. The results of both analyses are presented in Table 3. The proportion of Blacks in the sample is

1.2j

•

0.8 n

._N o41 ¢)

"

1

,

.

.

,

"

•

" .

0

0

,

.

"

"

"

.ec~*

• ,~.

•

o

:~"

.

?

• . ,,.,o..,

•

° ' oQ

: e

0•

~i~

"J

CO 13

-o,41

:,...

-0.81 -1.2

o.'.,

•

" ,

2

"..

•

,

. 6

significantly and substantially related to ES. For example, as the proportion of Blacks in a sample increases from .20 to .80, the ES would be expected to increase by .094 ( = [.80 - .20] x .157). In the second analysis, a multiple regression was used to examine racial consonance within a school. We combined two modera t o r s - t h e percentage of Black/White students in the school(s) from which Black/White participants were d r a w n - - i n a single regression model. For some ESs, each of one or more schools contributed both Black and White subsamples; for others, some or all of the schools contributed only one subsample. The regression coefficient estimate for each moderator (see Table 3) indicates the predicted ES increase for a one-percentage-point increment in the moderator (when the other moderator is controlled). The estimate was ~ = .310 for Black students and ~ = - . 3 0 2 for White students. This result suggests, for example, that an ES involving Black students from an 80% Black school would exceed one involving Black students from a 20% Black school by .186 ( = [.80 - .20] x t.3101) standard deviations. Similarly, the ES involving White students from an 80% White school would be lower than that involving White students from a 20% White school by .181 ( = [.80 - .20] x I-.3021) standard deviations. Thus, regression coefficient estimates for both moderators indicate that as the school environment becomes more consonant for a particular group, the self-esteem advantage for that group increases.

Instrumentation Effects ,

.

,

i 10 Age

,

,

,

, 14

.

.

•

i 18

,

,

, 22

(years)

Figure 1. Scatterplot of Hedges's d against average participant age in the study. Effect size (ES) weights are restricted to 700. Symbol area is proportional to ES weight. Dotted horizontal line represents no effect (d = 0.0). Solid line represents weighted linear regression equation for all independent ESs with available data (k = 257).

Standardization. On the basis of previous summaries and the assumption that standardized instruments detect existing differences more reliably, we expected racial comparisons derived from standardized instruments to yield larger ESs than would those from unstandardized instruments. Results from this analysis are pre7 In our notation, g is the estimated contrast.

36

GRAY-LI'~'LE AND HAFDAHL Table 4

Categorical Moderator Analyses: Participant Characteristics ~lo

Source/group Participant sex Total Between groups Within groups Female Male Socioeconomic status Total Between groups Within groups Low Middle High

SE(cl°)

95% CI for 8

.160

.013

(. 135,. 185)

.176 .141

.017 .019

(. 142, .210) (.105, .178)

.050

.010

(.031, .068)

.142 .016 - .021

.017 .014 .020

(.110, .175) (-.012, .044) (-.060, .018)

Q

df

592.00* * 1.85 590.15"* 291.73"* 298.42**

139 1 138 62 76

734.23** 49.66** 684.57** 272.09** 359.94** 52.53**

128 2 126 75 41 10

Note. do = pooled weighted effect size; SE((Io) = standard error of do; 95% CI for 8 = 95% confidence interval about the population effect size; Q = heterogeneity statistic (distributed approximately as chi-square) for the source or group, representing heterogeneity accounted for by the moderator (between groups), or between-effect sizes heterogeneity not explained by the moderator (within groups and within each group); df = degrees of freedom for Q. **p < .0001.

sented in Table 5. A test of between-group heterogeneity was significant. As predicted, a significantly higher positive mean ES (i.e., a Black self-esteem advantage) was obtained with standardized instruments. For unstandardized instruments, the average ES was substantially smaller but still positive. Comparison of individual instruments. The results of the comparisons of RSE, SEI, TSCS, and PH scales are presented in Table 5. All mean ESs were positive but varied in magnitude. The largest was found with the RSE; this is a moderate effect suggesting that self-esteem of Black respondents may exceed that of Whites by nearly one quarter of a standard deviation. Relatively smaller ESs were associated with use of the SEI and TSCS. Contrasts among the TSCS, SEI, and PH revealed no significant

0.5 .~ . . . .

19

.N 09

0.4

t

All female

25%male

0.3

50% male

ILl "O

0.1

All m

•~

o.o

"6

-o.1 13_

"" f

75% m a l e

#= (1)

j - ~ JJ .

"5

~

~

l?J

-0.2 -0.3

2

6

1'0

14

1'8

22

differences. A contrast of the RSE, the only global instrument, with the other three instruments combined revealed a nonsignificant trend, g = .224, X2(1) = 2.86, p = .091. Although suggestive, this result does not offer reliable evidence that a larger ES would be found with global than with aggregate instruments. Response format. We contrasted results from the two aggregate instruments (SEI and PH), whose true-false response format precludes extreme responding, with results from the TSCS, which is also an aggregate scale but with a Likert-type response format. The contrast of SEI and PH with the TSCS was not significant, p = .876. Moreover, a contrast of both scales having a Likert-type response format, RSE and TSCS, with the two true-false scales, PH and SEI, was also nonsignificant, p = .349. Thus, there was no support for the hypothesis that extreme responding is an important contributor to the finding of higher self-esteem in Black respondents. Domain. We expected a larger positive ES with general instruments than with specific measures of academic self-esteem. The results are presented in Table 5. The 217 comparisons using general, aggregate instruments yielded a significantly larger positive ES than did the 37 using academic self-esteem instruments, though the direction was the same for both types of instrument. The contrast between academic instruments and the RSE, which is entirely free of reference to content domains such as school or academics, was even larger, g = .211, X2(1) = 26.40, p < .001. Although illustrative of only one specific domain, these analyses support the hypothesis that racial differences in self-esteem vary with the domain assessed.

Age (years) Characteristics of the Authors and the Research Report Figure 2. Simple regression lines for Hedges's d regressed on average participant age for five different proportions of male participants in the sample. Effect size weights are restricted to 700. Dotted horizontal line represents no effect (d = 0.0). Solid line represents weighted linear regression for comparisons using only male participants; dashed line broken by two dots, only female participants.

Author race. This analysis was based on 100 comparisons in which author race was Black or White. Author race was known for 17 additional comparisons; however, no other ethnic categories were frequent enough to form additional groups. The average ESs obtained by Black and White investigators were in the same

37

RACIAL COMPARISONS OF SELF-ESTEEM Table 5

Categorical Moderator Analyses: Instrumentation Effects Source/group Instrument standardization Total Between groups Within groups Standardized Unstandardized Standardized instrument Total Between groups Within groups Coopersmith Piers-Harris Rosenberg Tennessee Instrument domain Total Between groups Within groups General Academic

tl°

SE(~lo)

95% CI for ~

Q

df

.148

.007

(. 135,. 161 )

.184 .124

.010 .008

(. 164, .204) (. 107,. 140)

1,673.72 * * 20.56** 1,653.16'* 753.40** 899.76**

260 1 259 176 83

.179

.016

(. 147, .210)

.154 .194 .246 .168

.026 .044 .039 .029

(.103, (.107, (. 169, (. 111,

386.46** 4.16 382.30** 161.99"* 54.01" 46.41" 119.90"*

110 3 107 36 25 17 29

.142

.007

(. 129,. 155)

.201 .035

.008 .011

(.184, .217) (.013, .058)

1,613.32"* 137.33"* 1,475.99"* 1,170.52"* 305.47 **

253 1 252 216 36

.204) .281) .324) .225)

Note. do = pooled weighted effect size; SE(~lo) = standard error of do; 95% CI for/~ = 95% confidence interval about the population effect size; Q = heterogeneity statistic (distributed approximately as chi-square) for the source or group, representing heterogeneity accounted for by the moderator (between groups), or between-effect sizes heterogeneity not explained by the moderator (within groups and within each group); df = degrees of freedom for Q. * p < .001. * * p < .0001.

direction, indicating higher self-esteem for Black respondents. The difference in ESs, though in the expected direction, was not significant (Table 6) and did not confirm the hypothesized relationship b e t w e e n ES and race o f the investigator. Year of publication. The association o f publication year with self-esteem differences was e x a m i n e d by a simple regression analysis (see Table 3). The association was in the expected positive direction but was not significant. Thus, the prediction that ES

would be larger in studies conducted in more recent years was not confirmed by this analysis. Publication outlet. This analysis examined ES differences associated with publication in journals versus dissertations. ESs from both sources were positive, but the average ES for dissertations was significantly higher (see Table 6). The more frequent use o f standardized measures by dissertation authors may be a partial contributor to this difference.

Table 6

Categorical Moderator Analyses: Author and Research Report Characteristics Source/group Author race Total Between groups Within groups Black White Publication outlet Total Between groups Within groups Journal/book Dissertation

clo

SE(~lo)

95% CI for ~

.089

.010

( .070,. 108 )

.104 .088

.034 .010

(.038, .170) (.068, .107)

.148

.007

(.135, .161)

.124 .212

.008 .012

(. 108,. 139) (. 188, .236)

Q

df

547.38 * * 0.21 547.17"* 79.36** 467.81"*

99 1 98 18 80

1,673.72"* 36.90** 1,636.83"* 1,163.08 * * 473.74" *

260 1 259 134 125

Note. do = pooled weighted effect size; SE(~Io) = standard error of~to; 95% CI for ~ = 95% confidence interval about the population effect size; Q = heterogeneity statistic (distributed approximately as chi-square) for the source or group, representing heterogeneity accounted for by the moderator (between groups), or between-effect sizes heterogeneity not explained by the moderator (within groups and within each group); df = degrees of freedom for Q. ** p < .0001.

38

GRAY-LITTLE AND HAFDAHL

Multiple-Moderawr Model In most of the previous analyses, moderators were examined separately. However, to represent relationships among factors moderating study outcomes, we attempted to build a multiplemoderator model. The initial model included three dummy-coded categorical moderators--instrument standardization and domain, and publication t y p e - - a n d three centered (weighted mean = 0) continuous moderators--age, proportion of male participants, and proportion Black in the sample. These moderators were available for enough comparisons to yield an adequate number of ESs with complete information (k = 225). (SES was significant but was not included because missing SES information would have eliminated more than half the ESs from this model.) The initial model included these six moderators and their 15 two-way interactions. We systematically dropped terms whose exclusion did not significantly increase residual heterogeneity; a nonsignificant main effect was retained if it was part of a significant interaction. This removal procedure resulted in a model with nine significant interactions; two were categorical-categorical, and seven were categorical-continuous. Because each main effect was involved in at least two interactions, all six were retained. Table 7 includes the regression coefficient estimate for each tenn. As all main effects were qualified by interactions, they are not discussed further, except to note that their directions in this model are similar to those in the single-moderator analyses (see Tables 3 through 6). The coefficient estimate of each interaction term can be interpreted as in ordinary multiple regression; that is, it indexes the degree to which the two moderators' effects are nonadditive, assuming that all first-order terms not in the interaction are held constant.

Continuous-categorical interactions. Each interaction between a continuous and a dichotomous moderator suggests that the two groups differ in the degree of the continuous moderator's association with ES. Participant age interacts with both standardization and domain: The greater Black self-esteem advantage associated with standardized instruments increases with age, whereas the advantage associated with general versus academic instruments decreases among older participants. Proportion of male participants yielded significant interactions with standardization and publication type: When standardized instruments are used, one would expect ESs to increase more rapidly (or decrease less rapidly) in increasingly male samples than when unstandardized instruments are used. Similarly, one would expect ESs reported in journals or books to decrease more rapidly than those from dissertations in increasingly male samples. Finally, the proportion of Blacks in the sample interacts with all three categorical moderators. Were the proportion Black to increase from zero to one, the Black selfesteem advantage associated with standardized instruments would increase, whereas the same advantage associated with measures of general self-esteem and with journal publications would decrease substantially. That is, whereas the effect (on ES) of standardization is greater for research using more Black as compared with White participants, the effects of instrument domain and of publication outlet are larger for research using more White participants. Categorical-categorical interactions. Both interactions between dichotomous moderators involve instrument domain. The significant interaction between standardization and domain (~ = - . 5 6 3 ) indicates that the mean ES difference between standardized and unstandardized instruments differs for measures of gen-

Table 7

Multiple-Moderator Analysis Moderator

~

SE(~)

95% CI for/3

Q,,sld

df

Participant age (A)~ P(male) (M)a'b P(Black) (B)~'b Standardization (S) c Domain (D)d Publication outlet (P)~ A x S A XD M x S M x P B XS B XD B XP S × D D XP Intercept

0.045 -0.086 0.184 0.511 0.113 -0.132 0.056 -0.072 0.233 -0.213 0.323 -0.386 -0.436 -0.563 0.234 -0.099

0.005 0.065 0.150 0.081 0.056 0.061 0.009 0.010 0.066 0.055 0.150 0.166 0.111 0.088 0.062 0.050

(.036, .054) (-.212, .041) (-.111, .479) (.352, .670) (.004, .222) (-.251, -.013) (.038, .074) (-.091,-.053) (.103, .363) (-.320,-.106) (.029, .617) (-.712, -.060) (-.653,-.219) (-.736, -.390) (.112, .356) ( - . 197, -.001)

783.16"*

209

Note. ~ = weighted least squares unstandardized regression coefficient estimate; SE(~) = standard error of ~; 95% CI for /3 = 95% confidence interval about the population regression coefficient; Q~ia = residual heterogeneity statistic (distributed approximately as chi-square), representing between-effect sizes heterogeneity not explained by the regression; df = degrees of freedom for Qresid; P ( ' ) = proportion of • in the sample or school(s). a Each continuous moderator's weighted mean over these 225 comparisons was centered at zero. b Proportion base rate is the sample of respondents included in the comparison, c Instrument standardization, dummy-coded as unstandardized = 0, standardized = 1. d Instrument domain, dummy-coded as academic = 0, general = 1. * Publication outlet, dummy-coded as dissertation = 0, journal/book = 1. • * p < .0001.

RACIAL COMPARISONS OF SELF-ESTEEM eral versus academic self-esteem. That is, the expected (standardized minus unstandardized) ES difference is .563 standard deviations lower when general rather than academic self-esteem instruments are used. The significant positive interaction between publication type and domain suggests that the mean ES difference between publication types (higher ESs for dissertations) increases about .234 standard deviations when general versus academic self-esteem is measured. The multiple-moderator analysis was exploratory, and its results should be interpreted cautiously in light of the weighted nature of the analysis, the unbalanced distribution of ESs among moderator categories, the heterogeneity left unexplained by this model, and missing data. It should also be noted, however, that results of the single- and multiple-moderator analyses are largely consistent.

Mixed-Effects Models The fixed-effects moderator models presented above left substantial ES heterogeneity unexplained. As this residual heterogeneity is not attributable to random error from sampling participants, it suggests that ES populations vary on attributes excluded from the model (Hedges, 1994; Shadish & Haddock, 1994). Although we interpreted estimates from these fixed-effects models, such models may be misspecified and may yield biased parameter estimates (e.g., subgroup mean ESs, regression coefficients). An alternative is a mixed-effects model, created by adding a random variance component that explicitly captures the variability of population parameters. (See Hedges & Vevea, 1998, and Overton, 1998, for discussions of differences among fixed-, random-, and mixed-effects models.) Mixed-effects models not only account for residual ES heterogeneity but also support the generalization of inferences to a larger universe of studies than those used in the sample. However, this increased generalizability comes at the cost of larger standard errors and impaired sensitivity to moderator effects. For each (set of) moderator(s) reported in Tables 3 through 7, a mixed-effects model was estimated using a maximum-likelihood estimation algorithm (Vevea & Hedges, 1995). Likelihood ratio tests were used to test for systematic moderator effects in the presence of a variance component. Several significant fixed-effects moderators became nonsignificant. Such nonsignificance can indicate that these moderators account for little true ES heterogeneity or, as we believe, that standard errors augmented by the variance component overwhelm even truly important moderators. Nevertheless, the single moderators participant age (p < .001), instrument domain (p = .015), specific standardized instrument (p = .010), and publication year (p = .032) did contribute explanatory power beyond the variance component, as did the Age × P(Male) interaction model (p < .001) and the multiplemoderator model (p < .001). The detection of these effects in the presence of unexplained random heterogeneity is evidence of their robustness and lends support to the findings of the fixed-effects models presented above.

39

ences in self-esteem that have been and could be conducted, we investigated the selective publication of research results. One graphical method for detecting publication bias is to plot individual study ES estimates against risk factors for selective publication. Sample size (i.e., the traditional funnel plot) is one such factor, in that studies with smaller sample sizes are more prone to exclusion during publication. However, Vevea and Hedges's (1995) suggestion to plot ES against conditional sampling variance-roughly the reciprocal of sample size--is particularly appropriate for the present data, where the few extremely large samples would obscure any selection pattern among the small studies, those most prone to publication bias. Selection bias often appears as asymmetry due to omitted high or low ESs or sparseness among small-sample studies with ESs near zero. The modified funnel plot in Figure 3 displays ES against conditional sampling variance, with the latter computed using the overall pooled ES estimate. Along the entire range of d, ESs are distributed fairly symmetrically about the mean weighted ES (indicated by the solid horizontal line), suggesting little evidence for any typical selection bias pattern operating in our sample. Although one ES stands out as an extreme observation (Spigner, 1986; d = -.916, conditional variance = .418), being a single point, it is probably not cause for concern; further examination revealed that it did not result from erroneous reporting or data coding. In summary, these ESs seem to be unaffected by selection bias, and the results of our analyses can safely be assumed to represent most research undertaken on this issue. Discussion Before we discuss the substantive findings, a number of methodological issues warrant further consideration. First, participants in most studies were drawn from school settings. The samples of children and young adolescents are likely to represent the general population in that age range. However, the young adult samples may be less representative because by early adulthood many potential participants have discontinued their formal education. Second, we excluded studies of middle-aged and older adults

1.2 0.8

.N_ 09 *6 ILl

**

.

.

IR L*. "***

04

!

-o.4.. -0.8

-1.2

**.~ f

'

•

"~

*

,

0.00

0.11

I

I

0.22

0.33

0.44

Conditional Variance

Potential Biases in the Retrieved Literature The failure to include studies with certain features can bias meta-analytic results. To determine whether the comparisons included in our sample reasonably represent studies of racial differ-

Figure 3. Modified funnel plot of Hedges's d against conditionalsampiing variance.Solidhorizontalline representsweightedaverage effect size (ES; de = .148). Conditionalsamplingvariancewas computed using this overall pooled ES estimate rather than each conditionalES.

40

GRAY-LITTLEAND HAFDAHL

because studies of race and self-esteem in adults are less frequent and the samples less representative of the populations in those age groups. Third, as the moderator variables we examined were necessarily those studied with frequency in the existing literature, our review could not be entirely systematic or comprehensive. We lacked the data to answer either the full range of questions stemruing from previous literature or others generated by our own analyses, and thus, many important questions remain unanswered. Fourth, though we cast a wide net, we have no doubt missed research relevant to this topic. We do not believe, however, that any category of research has been neglected systematically.

Substantive Issues: Racial Differences in Self-Esteem This review shows that despite substantial similarity, Black children, adolescents, and young adults have higher average selfesteem than their White counterparts. The Black self-esteem advantage is contrary to classical theorizing regarding the relationship between self-esteem and social status. Although the effect is small, we view it with some confidence because it is found with different types of instruments and appears to be stronger when more adequate instruments are used. In the sections to follow, the discussion of present findings and previous research is oriented to three questions: Why do Black Americans not have lower selfesteem than White Americans? Why do they often have higher self-esteem? What factors affect the magnitude, and sometimes the direction, of the difference found? Why do Black Americans not have lower self-esteem? Reflected appraisal and social comparison were the theoretical constructs most often invoked as the basis for predicting lower selfesteem for Black Americans. The validity of reflected appraisal in predicting a Black self-esteem disadvantage hinged on two important assumptions: (a) that individual self-esteem is sensitive to the appraisals of the larger society, and (b) that the larger White society constitutes significant others for Blacks. Both assumptions have been challenged (Shrauger & Schoeneman, 1979). First, do the appraisals of others directly affect individual self-esteem? Felson's (1993) research suggested that perceptions of others' appraisals are more related to self-esteem than are the actual appraisals, partially because actual appraisals may not be fully known. However, actual and perceived appraisals are often related. In the United States, for example, societal attitudes toward racial groups are widely known. Thus, if individual self-esteem is directly influenced by the attitudes of larger society, it is likely that Blacks would experience a negative impact. However, Black selfesteem levels have not responded in the predicted way. This failed expectation may be related to the second question: Does the larger society constitute a significant other for individual self-esteem? Current research and theory strongly suggest that significant others for personal self-esteem consist of primary groups or immediate reference groups, not the larger society (Cairns, 1990; Hoelter, 1981; H. W. Marsh, 1987, 1990; Oyserman & Markus, 1993). Cairns (1990) suggested that social projection--manifested in the belief that minority group members held the same low opinion of themselves that others held of them--led behavioral scientists to expect low self-esteem for minority group members. This logic also implied divergent processes: in-group evaluation as the basis of self-esteem in Whites but out-group evaluation as the basis of self-esteem for Blacks. Transferring the reference group for Black

Americans to the larger society changed the meaning of the significant other construct so that it no longer referred to the most personally relevant others. A reformulated version of reflected appraisal would suggest that significant others for Blacks and Whites comprise their primary, proximal social group. Future research in this area could address this question directly by asking participants to identify the individuals and groups whose views and evaluations are important to their self-esteem (Felson, 1993). Social comparison predictions regarding race and self-esteem also hinged on the assumption that Blacks feel demeaned by comparison to higher status Whites. Rosenberg and Simmons (1972) noted long ago, however, that Black adolescents in majority Black schools compared themselves with other Black adolescents, just as White students who make up a majority might be expected to make in-race comparisons. Our data show that a racially dissonant setting is associated with a self-esteem disadvantage for both Black and White respondents. We do not intend to suggest that social comparison and reflected appraisal are irrelevant to understanding Black self-esteem. On the contrary, we believe that both processes are germane to individual self-esteem and that they have been misapplied in some previous efforts to understand group differences in self-esteem. Thus, social comparison and reflected appraisal imply parallel processes, but often different social referents, for Blacks and Whites. Why do Black Americans have higher self-esteem ? In the past, high Black self-esteem has not been accepted as a valid phenomenon and, in the main, three types of arguments have been used to rationalize it: response style artifact, a group-by-item interaction favoring Black respondents, or defensive responding. The present data are most relevant to the response style argument advanced by Bachman and O'Malley (1984). Our contrast of Likert-type and dichotomous scale formats revealed no support for the position that higher Black scores were attributable to extreme responding. M. C. Taylor and Walsh (1979) addressed a related question, whether the standard deviation of self-esteem scores is larger for Blacks than for Whites, and found no support. In view of previous research and present findings that scores for Blacks are also higher on scales that do not permit extreme responding, an extremity response set does not offer a compelling explanation of higher self-esteem among Blacks. A second interpretation attributes the Black self-esteem advantage to a group-by-item content interaction. Hoelter's (1982) selective credulity hypothesis is an example of this approach. Selective credulity postulates racial differences in response to scale content, that is, that higher scores for Blacks do not indicate generally more positive self-esteem but the presence of scale content that favors Black respondents. Members of different racial and ethnic groups may indeed respond to items in a way that reflects group membership and adaptation to their group status, as well as the individual differences the items were intended to tap. This is apparently the case with gender. For example, numerous studies indicate that physical appearance heads the list as a domain that contributes to self-esteem in adolescence and may be more important in the self-esteem of girls than boys (Harter, 1990; Pliner, Chaiken, & Flett, 1990). Similarly, for a given instrument or domain, there may be group-by-item interactions that favor White respondents, whereas other instruments may contain items that elicit more positive responses from Blacks. In the present

RACIAL COMPARISONS OF SELF-ESTEEM study, the relevance of content can be seen in the significant difference in ESs obtained with general and academic instruments. A systematic evaluation of group differences in response to varied formats and contents would provide a direct answer to these questions. First, it is necessary to determine whether the factorial structure of self-esteem measures is the same for Blacks and Whites and whether there are different patterns of responding on an item-by-item basis. This approach could be expanded through the use of item response theoretic analysis to examine differential item functioning for racial groups on frequently used measures. Ultimately, it is important to determine not only whether Blacks and Whites have similar response patterns to groups of self-esteem test items but also whether self-esteem scores have the same implications for psychological functioning and behavior. (See Hui & Triandis, 1985, and Knight & Hill, 1998, for discussions of measurement equivalence across racial and ethnic groups.) Why then do Blacks, particularly adolescents and young adults, exhibit a self-esteem advantage? We subscribe to the social identity theory explanation presented in the introduction: In their quest for positive social identity, Blacks, a highly identifiable social group, emphasize their desirable distinctiveness. Is there support for two key elements of this formulation: that Blacks emphasize the distinctiveness of their racial/ethnic identity more than Whites and that such emphasis is associated with individual self-esteem? Although data collected for this review cannot answer these questions, existing research can be brought to bear on both issues. Most research participants in the relevant studies were adolescents and young adults; thus, the evidence is applicable to the age groups where our results reveal a Black self-esteem advantage. The first source of information is research on racial identity. In general, African Americans treat ethnic or racial group membership as a meaningful category distinction. Several lines of research justify the conclusion that among Blacks and other minority groups, in comparison with Whites, ethnic/racial identity is experienced more intensely, ethnicity is a more clearly articulated source of self-identification, and ethnicity is more likely to be mentioned spontaneously as a component of the self-concept (Gaines et al., 1997; Judd, Park, Ryan, Brauer, & Kraus, 1995; McGuire & McGuire, 1988; Phinney, 1989; Phinney & Alipuria, 1990; Rotheram-Borus, 1990; Smith, 1991). This research strongly suggests that race is a more distinctive source of identity for Blacks than for Whites. Judd et al. (1995) also found that among Black college students, the level of in-group preference, or ethnocentrism, was higher; that is, Black respondents evaluated in-group characteristics more positively than out-group characteristics, a pattern that was reversed for Whites. A second requirement of our formulation is that racial or ethnic identity be an important source of variance in the self-esteem of Blacks. (See critiques of racial identity constructs and instruments by Burlew & Smith, 1991; W. E. Cross, 1991; Rowley et al., 1998.) A positive relationship has been found between individual self-esteem and strength of ethnic identity among Blacks in numerous studies (e.g., Clark, 1985; Goodstein & Ponterotto, 1997; Lorenzo-Hernandez & OueUette, 1998; Munford, 1994; Parham & Helms, 1985; Paul & Fischer, 1980; Phinney, 1988; Phinney & Alipuria, 1990; Phinney & Chavira, 1992; Pyant & Yanico, 1991; Rotheram-Borus, 1989; White & Burke, 1987; Wright, 1985). A smaller number of studies report no reliable relationship (see, e.g., Clark, 1982; Hines &Berg-Cross, 1981; Houston, 1984; McAdoo,

41

1977; Rotheram-Borus, 1990; Rowley et al., 1998). Studies showing no association were conducted primarily with preschool or elementary school age participants, whereas those reporting a positive association were largely conducted with adolescents and young adults, the age groups where the Black self-esteem advantage most pertains. On balance, these studies support the contention that ethnic identity is an important correlate of individual self-esteem in Black Americans. Moreover, studies including White participants commonly find little or no relationship between self-esteem and ethnic identity for Whites or that the association is weaker than that found for Blacks and other minority respondents (see, e.g., Goodstein & Ponterotto, 1997; Phinney, 1988; Phinney & Alipuria, 1990). Thus, current evidence suggests both that ethnic identity is more salient for Blacks than for Whites and that ethnic identity is associated more strongly with individual self-esteem among Blacks. Further, when Judd et al. (1995) compared the attitudes of college students with those of adults (mean age = 46), they found that ethnicity was more salient in the identity of Black college students than among Black adults. For White Americans, however, the younger groups placed less conscious emphasis on racial or ethnic membership. Thus, the relationship of ethnicity to individual self-esteem seems to have diverged for the two racial groups. The social identity formulation is offered as an explanation of the Black self-esteem advantage rather than as a general explanation of majority-minoritydifferences. The directional difference in self-esteem between Whites and other ethnic minorities may vary (e.g., work by Gray-Little & Hafdahl, 1999, suggests that Hispanic groups report lower mean self-esteem than Whites). Moreover, the relationships of self-esteem to public and private evaluations of one's ethnic group may not be the same for all ethnic minority groups. For example, Crocker, Luhtanen, Blaine, and Broadnax (1994) compared three ethnic groups on the relationship between their views of their own group and the way that they perceived that the group was perceived by others. For Asian Americans, the relationship was quite strong; for Whites, moderate; but for Blacks, near zero. The diverse patterns of association may reflect cultural differences within the three groups and their specific histories within this country, but they also suggest using caution in equating the meaning of ethnicity across groups (Porter & Washington, 1993).

Characteristics Affecting Racial Difference in Self-Esteem Although we found Black Americans to have a slight selfesteem advantage, it is not difficult to imagine a combination of factors--such as a preponderance of samples with younger, high SES students tested with unstandardized instruments--that might yield different results. Among the moderators studied, the magnitude of racial differences in self-esteem was associated primarily with participant characteristics and with instrumentation. Participants and their social context. The self-esteem advantage for Black respondents increases with age. (Indeed, before preadolescence, there appears to be a White self-esteem advantage.) This moderately large, robust effect is present even when other sample and instrumentationcharacteristics are controlled and when a mixed-effects model is used. Increased self-esteem during adolescence has been attributed to increased motivation for selfenhancement, more freedom to select a peer group with similar

42

GRAY-LITTLEAND HAFDAHL

values and interests, and greater ability to behave in socially reinforced ways (McCarthy & Hoge, 1982; Rosenberg, 1986). A number of theorists have also speculated that in adolescence the bases for self-esteem become more internalized, abstract, and differentiated (Cairns, 1990; Demo & Savin-Williams, 1992; Rosenberg, 1986). However, none of these explanations has been fully tested by research, nor do they explicate racial differences in the development of self-esteem. Given that children's awareness of racial prejudice and of the implications of minority group status increases withage, one might reasonably predict decrements with age in African American self-esteem relative to Whites. We suspect that the age effect is attributable partially to the differential ethnic identification among Black and White adolescents and to the fact that ethnic identity is more germane to self-esteem in Blacks. The larger Black self-esteem advantage among adolescent and young adult samples with a larger proportion of female participants may also be related to combined differences in ethnic identification and in the socialization of gender roles. A full explanation of the age × sex interaction is beyond the scope of this article; however, a prototype for such an explanation might focus on physical attractiveness. For example, adolescent girls often rate themselves lower in physical attractiveness than do boys (see, e.g., Harter, 1990; Pliner et al., 1990). This gender difference appears more pronounced among Whites than among Blacks (Simmons & Blyth, 1987; Tashakkori, 1993). Some evidence also suggests that attractiveness ratings are higher among Black than White adolescents (see, e.g., Pallas, Entwisle, & Alexander, 1990; Tashakkori, 1993). In combination, these trends would lead to a depression of self-esteem scores for girls relative to boys at adolescence, when concern with appearance is magnified, but to a greater change for White girls. Obviously, longitudinal research is needed to study the ways in which race and gender interact with immediate contextual variables, such as peer or family relations, or with individual characteristics, such as self-perceived physical attractiveness, to affect the development of self-esteem from childhood, through adolescence, and into adulthood (S. E. Cross & Madson, 1997; Oyserman & Markus, 1993). The social make-up of the school settings from which respondents were drawn was related to ES. Both Black and White students experienced a self-esteem advantage in racially consonant environments. Students who are in the majority have greater opportunities for friendship and social support, a sense of belonging, and interaction with others sharing the same value system, all of which can enhance self-esteem. The self-esteem advantage for Whites in the high SES category may also reflect the influence of social consonance. (Because of the potentially strong association between SES and racial composition, however, it is difficult to assess the unique effects of either.) Thus, a more general interpretation is that for both Blacks and Whites, self-esteem is likely to be higher in settings that are consonant for them; such consonance might be a function of race and other characteristics such as SES (see Gray-Little & Carels, 1997, for a discussion of self-esteem and consonance). Instrumentation. The finding that larger positive effects were obtained with standardized instruments is consistent with Wylie's (1979) observation that the methodologically adequate studies were more likely to show higher self-esteem among Blacks. We interpret the present finding simply as demonstrating that more

reliable and valid instruments are more likely to reveal differences if they are present. The Black self-esteem advantage was larger for general self-esteem instruments than for measures of academic self-esteem; however, the ES for both types of instruments was positive. This finding is of interest for both what it did and what it did not reveal. First, this analysis suggests that the extent of racial differences found may vary with the aspect of self-esteem being measured. Second, although past research provided evidence that academic self-esteem covaries with academic achievement and that the academic achievement of Whites is higher than that of Blacks, this analysis did not indicate that Whites had higher academic self-esteem than Blacks. One implication of this finding is that the relative contribution of academic achievement to academic self-esteem may differ for the two groups (Demo and Parker, 1987; Mboya, 1988). Attenuation of the academic performance-self-esteem relationship among Black students has been variously attributed to "academic disidentification" (J. W. Osborne, 1995; Steele, 1997), cultural values oppositional to school achievement (Boykin, 1994; Ogbu, 1994), and selectivity in values or in the choice of reference group (Crocker & Major, 1989; Rosenberg & Simmons, 1972). These concepts all suggest that Black students identify less than White students with the values and goals of educational institutions, resulting in a low correlation between their objective performance and subjective evaluation. Unfortunately, the limitations of our data do not allow us to distinguish the extent to which our findings regarding domain and racial differences are attributable to these or other mechanisms. Our findings do suggest, however, that the personal relevance and importance of a domain of experience are critically important in understanding not only individual self-esteem, as proposed by James (1890), but also group differences in that domain. Final Comments As we reviewed the variety of findings generated by the questions under consideration, we were reminded of Kagan's (1988) admonition regarding the importance of context: "The theoretical meaning of propositions about psychological processes, and therefore, their truth value, requires specification of a class of organism and context. . . . [M]any psychological qualities do not have the same meaning across different agents" (p. 618). We have found that the use of a specific procedure (e.g., a standardized vs. an unstandardized instrument) and a sample with specific characteristics (e.g., preschool children vs. adolescents) or a given racial composition represent particular contexts for studying self-esteem. Each context lends a particular meaning to racial comparisons. These results also suggest that the valid application of theoretical ideas such as reflected appraisal or social comparison might depend on whether the level of analysis is the individual or group. An important conceptual challenge in characterizing our findings stems from another fundamental problem: difficulties inherent in using racial membership as an independent variable. No doubt, it is the seeming simplicity of race, its social significance, and its permanence that seduce investigators into considering it an important independent variable when studying self-esteem or other personality processes. Race provides a convenient, salient social manipulation. However, racial group membership is confounded with culture, minority status, social class, and other variables, each of which may have implications for self-esteem. One reaction to

RACIAL COMPARISONS OF SELF-ESTEEM the complexity in conceptualizing race and ethnicity is to recommend that comparative research not be done, partially because it has often reinforced stereotypes and frequently ignores more important and more substantial within-group variations. However, an important contribution of such research is that it cautions researchers with regard to the limitations of theories concerning the relationship between social status (racial/ethnic minority) and individual psychological processes. Thus, comparative research can be helpful in refining theory. Rather than abandon research based on social categories (culture, ethnicity, gender, or race), researchers should attempt to understand the meaning of the categories in social and psychological terms.

References References marked with an asterisk indicate studies included in the meta-analysis. *Abbott, A. A. (1981). Factors related to third grade achievement, selfperception, classroom composition, sex, and race. Contemporary Educational Psychology, 6, 167-179. *Ahr, T. J. (1989). The effects of ability grouping on academic achievement and self-concept among Black and White students (Doctoral dissertation, University of Georgia, 1989). Dissertation Abstracts International 50(6A), 1556-1557. Anastasi, A., & Urbina, S. (1997). Psychological testing (7th ed.). Upper Saddle River, NJ: Prentice Hall. Anderson, N. B. (1989). Racial differences in stress-induced cardiovascular reactivity and hypertension: Current status and substantive issues. Psychological Bulletin, 105, 89-105. *Atolagbe, E. O. (1975). Correlation of self-concept and values in social classes, races and sexes (Doctoral dissertation, Ohio State University, 1975). Dissertation Abstracts International, 36(3-B), 1399-1400. *Bachman, J. G. (1970). Youth in transition (Vol. 2). Ann Arbor, Mh Institute for Social Research. Bachman, J. G., & O'Malley, P. (1984). Black-White differences in self-esteem: Are they affected by response styles? American Journal of Sociology, 90, 624-639. Baldwin, J. A. (1979). Theory and research concerning the notion of Black self-hatred: A review and reinterpretation. Journal of Black Psychology, 5, 51-77. Ballard, B., & Keller, H. R. (1976). Development of racial awareness: Task consistency, reliability, and validity. Journal of Genetic Psychology, 129, 3 - l l . Banaji, M. R., & Prentice, D. A. (1994). The self in social contexts. Annual Review of Psychology, 45, 297-332. Banks, J. (1984). Black youths in predominantly White suburbs: An exploratory study of their attitudes and self-concepts. Journal of Negro Education, 53, 3-17. Banks, W. C. (1976). White preference in Blacks: A paradigm in search of a phenomenon. Psychological Bulletin, 83, 179-186. *Barnes, M. E., & Farrier, S. C. (1985). A longitudinal study of the self-concept of low income youth. Adolescence, 20, 199-205. Baumeister, R. F., & Leary, M. R. (1998). The need to belong: Desire for interpersonal attachments as a fundamental human motivation. Psychological Bulletin, 117, 497-529. Baumeister, R. F., Smart, L., & Boden, J. M. (1996). The relation of threatened egotism to violence and aggression: The dark side of high self-esteem. Psychological Review, 103, 5-33. *Bennett, P. D., & Lundgren, D. C. (1976). Racial composition of day care centers and the racial attitudes and self-concepts of young Black and White children. Journal of Intergroup Relations, 5, 3-14. *Benson, J., & Rentsch, J. (1988). Testing the dimensionality of the

43

Piers-Harris Children's Self-Concept Scale. Educational and Psychological Measurement, 48, 615-626. *Blackboum, J. M., & Blackbourn, V. (1987). Self-concepts of young handicapped children: An analysis of race and sex. Perceptual & Motor Skills, 65, 626. Blascovich, J., & Tomaka, J. (1991). Measures of self-esteem. In J. P. Robinson, P. R. Shaver, & L. S. Wrightsman (Eds.), Measures of personality and social psychological attitudes (Vol. 1, pp. 115-160). San Diego, CA: Academic Press. *Bowler, R., Rauch, S., & Schwarzer, R. (1986). Self-esteem and interracial attitudes in Black high school students: A comparison with five other ethnic groups. Urban Education, 21, 3-19. Boykin, A. W. (1994). Harvesting talent and culture: African-American children and educational reform. In R. Rossi (Ed.), Schools and students at risk (pp. 116-138). New York: Teachers College Press. *Brack, C. J., Orr, D. P., & Ingersoll, G. (1988). Pubertal maturation and adolescent self-esteem. Journal of Adolescent Health Care, 9, 280-285. Brand, E. S., Ruiz, R. A., & Padilla, E. M. (1974). Ethnic identification and preference: A review. Psychological Bulletin, 81, 860-890, Brewer, M. B. (1991). The social self: On being the same and different at the same time. Personality and Social Psychology Bulletin, 17, 475482. Brewer, M. B. (1993). The role of distinctiveness in social identity and group behavior. In M. Hogg & D. Abrams (Eds.), Group motivation. London: Harvester-Wheatsbeaf. *Bridgette, R. E. (1970). Self-esteem in Negro and White southern adolescents (Doctoral dissertation, University of North Carolina, 1970). Dissertation Abstracts International, 31, 2977B. *Brnch, M. A., Kunce, J., & Egeman, D. F. (1972). Parental evaluation: A protection of self-esteem. Journal of Counseling Psychology, 19, 555558. *Burbach, H. J., & Bridgeman, B. (1976). Relationship of self-esteem and locus of control in Black and White fifth-grade students. Child Study Journal, 6, 33-37. *Burger, M. L. (1973). A comparative study of self-esteem among young Black, Spanish, and White children (Doctoral dissertation, Northern Illinois University, 1973). Dissertation Abstracts International, 34(4-A), 1689. *Burke, P. J., & Hoelter, J. W. (1988). Identity and sex-race differences in educational occupational aspirations formation. Social Science Research, 17, 29-47. Burlew, A. K., & Smith, L. R. (1991). Measures of racial identity: An overview and a proposed framework. Journal of Black Psychology, 17, 53-71. Bushman, B. J. (1994). Vote-counting procedure in meta-analysis. In H. Cooper & L. V. Hedges (Eds.), Handbook of research synthesis (pp. 193-213). New York: Russell Sage Foundation. Cairns, R. B. (1990). Developmental epistemology and self-knowledge: Towards a reinterpretation of self-esteem. In G. Greenberg & E. Tobach (Eds.), Theories of the evolution of knowing (pp. 69-86). Hillsdale, NJ: Erlbaum. *Carpenter, T. R., & Busse, T. V. (1969). Development of self-concept in Negro and White welfare children. Child Development, 40, 935-939. Cartwright, D. (1950). Emotional dimensions of group life. In M. L. Reymert (Ed.), Feelings and emotions (pp. 437-447). New York: McGraw-Hill. Caspi, A. (1987). Personality in the life course. Journal of Personality and Social Psychology, 53, 1203-1213. *Castillo, J. E. (1983). The relationship of minority identification, cognitive style, self-esteem, and other selected factors related to academic success in higher education (Doctoral dissertation, University of Michigan, 1983). Dissertation Abstracts International, 44(10-A), 3003. *Cicirelli, V. G. (1977). Relationship of socioeconomic status and ethnicity

44

GRAY-LITtLE AND HAFDAHL

to primary grade children's self-concept. Psychology in the Schools, 14, 213-215. Clark, M. L. (1982). Racial group concept and self-esteem in Black children. Journal of Black Psychology, 8, 75-88. Clark, M. L. (1985). Social stereotypes and self-concept in Black and White college students. Journal of Social Psychology, 125, 753-760. Coleman, J. S. (1966). Equality of educational opportunity. Washington, DC: U.S. Government Printing Office. *Coleman, V. D. (1979). The self-esteem and vocational maturity of females (Doctoral dissertation, Rutgers University, New Brunswick, 1979). Dissertation Abstracts International, 40(7-A), 3883. *Collins, M. A. (1993). School social context, self-esteem, and locus-ofcontrol among White, Black, and Hispanic youth (Doctoral dissertation, University of Maryland College Park, 1992). Dissertation Abstracts International, 54(2-A), 692-693. Cooley, C. H. (1902). Human nature and the social order. New York: Scribner. Coopersmith, S. (1967). The antecedents of self-esteem. San Francisco: Freeman. *Cornell, D. G., Delcourt, M. A. B., Goldberg, M. D., & Bland, L. C. (1995). Achievement and self-concept of minority students in elementary school gifted programs. Journal of Education of the Gifted, 18, 189-207. *Cozart, D. C. (1989). The relationship among self-concept, race, socioeconomic status, and mathematics achievement in Black and White fifth grade students (Doctoral dissertation, Fairleigh Dickinson University, 1988). Dissertation Abstracts International, 49(7-A), 1758. *Crain, R. M., & Bracken, B. A. (1994). Age, race, and gender differences in child and adolescent self-concept: Evidence from a behavioralacquisition, context-dependent model. School Psychology Review, 23, 496-511. *Crocker, J., Luhtanen, R., Blaine, B., & Broadnax, S. (1994). Collective self-esteem and psychological well-being among White, Black, and Asian college students. Personality and Social Psychology Bulletin, 20, 503-513. Crocker, J., & Major, B. (1989). Social stigma and self-esteem: The self-protective properties of stigma. Psychological Review, 96, 608630. Cross, S. E., & Madson, L. (1997). Models of the self: Self-construals and gender. Psychological Bulletin, 122, 5-37. Cross, W. E. (1987). A two-factor theory of Black identity: Implications for the study of identity development in minority children. In M. J. Rotheram & J. S. Phinney (Eds.), Children's ethnic socialization (pp. 117-133). Newbury Park, CA: Sage. Cross, W. E. (1991). Rethinking Nigrescence. In W. E. Cross (Ed.), Shades of Black: Diversity in African American identity (pp. 189-217). Philadelphia: Temple University Press. *Dales, R. J., & Keller, J. F. (1972). Self-concept scores among Black and White culturally deprived adolescent males. Journal of Negro Education, 41, 31-34. *Davids, A. (1973). Self-concept and mother-concept in Black and White pre-school children. Child Psychiatry and Human Development, 40, 30-43. *Davidson, H. R. (1974). Personality characteristics of Black high school students: An empirical and theoretical examination of interdisciplinary thought (Doctoral dissertation, California School of Professional Psychology, 1974). Dissertation Abstracts International, 35(10-B), 5104. Deaux, K. (1993). Reconstructing social identity. Personality and Social Psychology Bulletin, 19, 4-12. Demo, R. M. (1985). The measurement of self-esteem: Refining our methods. Journal of Personality and Social Psychology, 48, 1490-1502. *Demo, R. M., & Parker, K. D. (1987). Academic achievement and self-esteem among Black and White college students. Journal of Social Psychology, 127, 345-355.

Demo, R. M., & Savin-Williams, R. C. (1983). Early adolescent selfesteem as a function of social class: Rosenberg and Pearlin revisited. American Journal of Sociology, 88, 763-774. Demo, R. M., & Savin-Williams, R. C. (1992). Self-concept stability and change during adolescence. In T. M. Brinthaupt (Ed.), Self-perspectives across the life span (pp. 116-148). Albany, NY: State University of New York Press. Duncan, O. D. (1961 ). A socioeconomic index for all occupations. In A. J. Reiss, Jr. (Ed.), Occupations and social status (pp. 109-138). New York: Free Press. *Edwards, C. D. (1972). Stress in the school: A study of anxiety and self-esteem in Black and White elementary school children (Doctoral dissertation, Florida State University, 1972). Dissertation Abstracts International, 33(6-B), 2806. *Edwards, D. W. (1974). Blacks versus Whites: When is race a relevant variable? Journal of Personality and Social Psychology, 29, 39-49. *Entwisle, D. R., Alexander, K. L., Pallas, A. M., & Cadigan, D. (1987). The emergent academic self-image of first graders: Its response to social structure. Child Development, 58, 1190-1206. Epps, E. G. (1975). Impact of school desegregation on aspirations, selfconcepts and other aspects of personality. Law and Contemporary Problems, 39, 300-313. Felson, R. B. (1993). The (somewhat) social self: How others affect self-appraisals. In J. Suls (Ed.), Psychological perspectives on the self (Vol. 4, pp. 1-26). Hillsdale, NJ: Erlbanm. Finkelstein, N. W., & Haskins, R. (1983). Kindergarten children prefer same-color peers. Child Development, 54, 502-508. Fishbein, M., & Azjen, I. (1975). Belief attitude, intention, and behavior. Reading, MA: Addison-Wesley. Fitts, W. H. (1965). Tennessee Self-Concept Scale manual Nashville, TN: Counselor Recordings and Tests. *Franklin, A. W. (1985). An assessment of assertiveness and self-esteem in female college students: A comparison of race and sex-role (Doctoral dissertation, Emory University, 1985). Dissertation Abstracts International, 46(11-By, 4011-4012. *Fu, V. R., Hinkle, D. E., & Korslund, M. K. (1983). A developmental study of ethnic self-concept among preadolescent girls. Journal of Genetic Psychology, 142, 67-73. Gaines, S. O., Jr., Marelich, W. D., Bledsoe, K. L., Steers, W. N., Henderson, M. C., Granrose, C. S., Barajas, L., Hicks, D., Lyde, M., Takahashi, Y., Yum, N., Rios, D. L., Garcia, B. F., Farris, K. R., & Page, M. S. (1997). Links between race/ethnicity and cultural values as mediated by racial/ethnic identity and moderated by gender. Journal of Personality and Social Psychology, 72, 1460-1476. *Gerken, K. C., Allen, M., & Snider, B. C. (1984). The relationship between voluntary reduction of minority isolation and varied measures of affect. Child Study Journal, 13, 277-295. Gerth, H., & Mills, C. W. (1953). Character and social structure: The psychology of social institutions. New York: Harcourt Brace. Goodstein, R., & Ponterotto, J. G. (1997). Racial and ethnic identity: Their relationship to and their contribution to self-esteem. Journal of Black Psychology, 23, 275-292. GoPaul-McNicol, S. (1988). Racial identification and racial preference of Black preschool children in New York and Trinidad. Journal of Black Psychology, 14, 65-68. Graham, S. (1994). Motivation in African Americans. Review of Educational Research, 64, 55-117. Gray-Little, B. (1980). Race and inequity. Journal of Applied Social Psychology, 10, 468-481. *Gray-Little, B., & Appelbaum, M. I. (1979). Instrumentality effects in the assessment of racial differences in self-esteem. Journal of Personality and Social Psychology, 37, 1221-1229. Gray-Little, B., & Carels, R. (1997). The effects of racial and socioeconomic consonance on self-esteem and achievement in elementary, junior

RACIAL COMPARISONS OF SELF-ESTEEM high, and high school students. Journal of Research on Adolescence, 7, 109-131. Gray-Little, B., & Duley, J. F. (1993). Black self-esteem: A half-century of inquiry. Unpublished manuscript, University of North Carolina at Chapel Hill. Gray-Little, B., & Hafdahl, A. R. (1999). The meaning of self-esteem across different ethnic groups. Unpublished manuscript, University of North Carolina at Chapel Hill. Gray-Little, B., & Stem, K. (1992). A review of the doll studies. Unpublished manuscript, University of North Carolina at Chapel Hill. Gray-Little, B., & Teddlie, C. B. (1978). Racial differences in children's responses to inequity. Journal of Applied Social Psychology, 8, 107116. Gray-Little, B., Williams, V. S. L., & Hancock, T. D. (1997). An item response theory analysis of the Rosenberg Self-Esteem Scale. Personality and Social Psychology Bulletin, 23, 443-451. Greenwald, A. G., & Schuh, E. S. (1994). An ethnic bias in scientific citations. European Journal of Social Psychology, 24, 623-639. *Gregory, R. D. (1977). Self-concept across sex, color, and teacher estimated annual family income level for students in grades seven through twelve (Doctoral dissertation, Duke University, 1977). Dissertation Abstracts International, 38(4-B), 1859. *Griffith, D. D. (1985). The relationship of integration/segregation, race, socioeconomic status, and locus of control upon sophomore students' self-concept and achievement (Doctoral dissertation, University of Akron, 1985). Dissertation Abstracts International, 46(1B), 289. *Grossman, B. (1981). Ethnic identity and self-esteem: A study of Anglo, Chicano, and Black adolescents in Texas (Doctoral dissertation, New School for Social Research, 1981). Dissertation Abstracts International, 42(08B), 3423. Gurin, P., Gurin, G., Lao, R., & Beattie, M. (1969). Internal-external control in the motivational dynamics of Negro youth. Journal of Social Issues, 25, 29-54. Hansford, B. C., & Hattie, J. A. (1982). The relationship between self and achievement/performance measures. Review of Educational Research, 52, 123-142. *Hare, B. R. (1980). Self-perception and academic achievement: Variations in a desegregated setting. American Journal of Psychiatry, 137, 683-389. Hare, B. R. (1985). Stability and change in self-perception and achievement among Black adolescents: A longitudinal study. Journal of Black Psychology, 11, 29-42. *Harris, A. R., & Stokes, R. (1978). Race, self-evaluation and the protestant ethic. Social Problems, 26(1), 71-85. Harter, S. (1982). The Perceived Competence Scale for Children. Child Development, 53, 87-97. Hatter, S. (1985a). Competence as dimension of self-evaluation: Toward a comprehensive model of self-worth. In R. L. Leahy (Ed.), The development of the self(pp. 55-121). Orlando, FL: Academic Press. Harter, S. (1985b). Manual for the Self-Perception Profile for Children (revision of the Perceived Competence Scale for Children). Denver, CO: University of Denver. Harter, S. (1990). Adolescent self and identity development. In S. S. Feldman & G. R. Elliot (Eds.), At the threshold: The developing adolescent (pp. 352-387). Boston: Harvard University Press. *Hartnagel, T. F. (1970). Father absence and self-conception among lower class White and Negro boys. Social Problems, 18, 152-163. *Healey, G. W., & DeBlassie, R. R. (1974). A comparison of Negro, Anglo, and Spanish-American adolescents' self-concepts. Adolescence, 9, 15-24. Hedges, L. V. (1994). Fixed effect models. In H. Cooper & L. V. Hedges (Eds.), Handbook of research synthesis (pp. 301-321). New York: Russell Sage Foundation. Hedges, L. V., & Becker, B. J. (1986). Statistical methods in the meta-

45

analysis of research on gender differences. In J. S. Hyde & M. C. Linn (Eds.), The psychology of gender: Advances through meta-analysis (pp. 14-50). Baltimore: John Hopkins University Press. Hedges, L. V., & Olkin, I, (1985). Statistical methods for meta-analysis. New York: Academic Press. Hedges, L. V., & Vevea, J. L. (1998). Fixed- and random-effects models in meta-analysis. Psychological Methods, 3, 486-504. Heiss, J., & Owens, S. (1972). Self evaluations of Blacks and Whites. American Journal of Sociology, 78, 360-370. *Henderson, D. G. (1977). Self-concepts of fifth and sixth grade students based on the race and sex of their teachers (Doctoral dissertation, Louisiana State University and A&M College, 1977). Dissertation Abstracts International, 38(12-A), 7138. *Hemandez, M. (1984). A study of children's racial attitudes and selfesteem (Doctoral dissertation, New York University, 1984). Dissertation Abstracts International, 45(6-B), 1914. *Hines, P., & Berg-Cross, L. (1981). Racial differences in global selfesteem. Journal of Social Psychology, 113, 271-281. Hoelter, J. W. (1981). Differential impact of significant others on selfevaluation: A multiple group analysis (Doctoral dissertation, Indiana University, 1981). Dissertation Abstracts International, 42(9-A), 0093. *Hoelter, J. W. (1982). Race differences in selective credulity and selfesteem. Sociological Quarterly, 23, 527-537. Hoge, D. R., Smit, E. K., & Hanson, S. L. (1990). School experiences predicting changes in self-esteem of sixth and seventh grade students. Journal of Educational Psychology, 82, 117-127. *Holaday, M., Callahan, K., Fabre, L., & Hall, C. (1996). A comparison of culture-free self-esteem scale means from different child and adolescent groups. Journal of Personality Assessment, 66, 540-554. Hollingshead, A. B., & Redlich, F. F. (1957). Two-factor index of social position. New Haven: Yale Station. *House, J. D. (1993). Achievement-related expectancies, academic selfconcept, and mathematics performance of academically underprepared adolescent students. Journal of Genetic Psychology, 154, 61-71. *Housely, K., Martin, S., McCoy, H., Greenhouse, P., Stigger, R., & Chopin, L. (1987). Self-esteem of adolescent females as related to race, economic status, and area of residence. Perceptual and Motor Skills, 64, 559 -566. Houston, L. (1984). Black consciousness and self-esteem. Journal of Black Psychology, 11, 1-7. Hraba, J., & Grant, G. (1970). Black is beautiful: A re-examination of racial preference and identification. Journal of Personality and Social Psychology, 16, 398-402. Hughes, M., & Demo, D. H. (1989). Self-perception of Black Americans: Self-esteem and personal efficacy. American Journal of Sociology, 95, 132-157. Hui, C. H., & Triandis, H. C. (1985). Measurement in cross-cultural psychology: A review and comparison of strategies. Journal of CrossCultural Psychology, 16, 131-152. *Hulbary, W. E. (1975). Race, deprivation and adolescent self-images. Social Science Quarterly, 56, 105-114. *Hunt, J. G., & Hunt, L. L. (1977). Racial inequality and self-image: Identity maintenance as identity diffusion. Sociology and Social Research, 61, 539-559. *Hunt, J. P. (1991). Self-concept comparisons among Black, MexicanAmerican, and White children in low-SES integrated schools (Doctoral dissertation, Arizona State University, 1991). Dissertation Abstracts International, 53(2A), 400. *Hurley, R. B. (1973). Race, fatherlessness and vocational development (Doctoral dissertation, New York University, 1972). Dissertation Abstracts International, 33(11-A), 6090-6091. James, W. (1890). The principles of psychology (Vol. 1). New York: Holt. *Jameson, A. S. (1973). An analysis of self-esteem and academic achievement of tri-racial isolate, Negro and Caucasian elementary and middle

46

GRAY-LII"I'LE AND HAFDAHL

school boys and girls (Doctoral dissertation, University of Maryland, 1973). Dissertation Abstracts International, 35(2-A), 722. Jensen, G. F., White, C. S., & Galliher, J. M. (1982). Ethnic status and adolescent self-evaluations: An extension of research on minority selfesteem. Social Problems, 30, 226-239. Jessor, R., & Jessor, S. L. (1977). Problem behavior and psychosocial development: A longitudinal study of youth. New York: Academic Press. *Jones, C. H. (1973). The relationships of self-esteem, general anxiety and test anxiety in Black and White elementary school students in grades four through six (Doctoral dissertation, Northern Illinois University, 1973). Dissertation Abstracts International, 34(3-A), 1131. *Jones, E. E. (1979). Personality characteristics of Black youth: A crosscultural comparison. Journal of Youth and Adolescence, 8, 149-159. Josephs, R. A., Markus, H. R., & Tafarodi, R. W. (1992). Gender and self-esteem. Journal of Personality and Social Psychology, 63, 391402. Judd, C. M., Park, B., Ryan, C. S., Brauer, M., & Kraus, S. (1995). Stereotypes and ethnocentrism: Diverging interethnic perceptions of African American and White American youth. Journal of Personality and Social Psychology 69, 460-481. Kagan, J. (1988). The meaning of personality predicates. American Psychologist, 88, 614-629. *Kearney, A. G. (1973). Factors affecting the development of self-esteem in young Black children (Doctoral dissertation, Rutgers State University, 1973). Dissertation Abstracts International, 34(7-A), 3874. *Keith, T. Z., Pottebaum, S. M., & Eberhart, S. (1986). Effects of selfconcept and locus of control on academic achievement: A large sample path analysis. Journal of Psychoeducational Assessment, 4, 61-72. Knight, G+ P., & Hill, N. E. (1998). Measurement equivalence in research involving minority adolescence. In V. C. McLoyd & L. Steinberg (Eds.), Studying minority adolescents (pp. 183-210). Mahwah, NJ: Erlbaum. *Kohr, R. L., Coldiron, J. R., Skiffington, E. W., Masters, J. R., & Blust, R. S. (1988). The influence of race, class, and gender on self-esteem for fifth, eighth, and eleventh grade students in Pennsylvania schools. Journal of Negro Education, 57, 467-481. *Kugle, C. L., Clements, R. O., & Powell, P. M. (1983). Level and stability of self-esteem in relation to academic behavior of second graders. Journal of Personality and Social Psychology, 44, 201-207. *Kuhn, K. C. (1973). Self-concept of Black and White students between and among social classes in newly desegregated elementary schools (Doctoral dissertation, Syracuse University, 1972). Dissertation Abstracts International, 33(9-A), 4945-4946. *Laryea, E. B. (1972). Race, self-concept and achievement (Doctoral dissertation, Columbia University, 1972), Dissertation Abstracts International, 33(5-A), 2172-2173. *Lawrence, W., & Brown, D. (1976). An investigation of intelligence, self-concept, socioeconomic status, race, and sex as predictors of career maturity. Journal of Vocational Behavior, 9, 43-52. *Lay, R., & Wakstein, J. (1985). Race, academic achievement, and selfconcept of ability. Research in Higher Education, 22, 43-64. *Lefebvre, A. (1973). Self-concept of American Negro and White children+ Acta Psychologica Taiwanica, 15, 25-30. *Leonards, J. S. (1988). An investigation of the joint relations of gender, ethnicity, and maturation level on factors of academic self-concept for elementary and secondary students (Doctoral dissertation, University of Southern California, 1987). Dissertation Abstracts International, 49(2A), 221. *Logan, R. C. (1984). Self-concept and self-actualization among Black and White college students (Doctoral dissertation, University of Pittsburgh, 1984). Dissertation Abstracts International, 45(5-A), 1345. *Long, B. H., & Henderson, E. H. (1968). Self-social concepts of disadvantaged school beginners. Journal of Genetic Psychology, 113, 41-51. *Long, B. H., & Henderson, E. H. (1970). Social schemata of school

beginners: Some demographic correlates. Merrill-Palmer Quarterly, 16, 305-324. Lorenzo-Hemandez, J., & Ouellette, S. (1998). Ethnic identity, selfesteem, and values in Dominicans, Puerto Ricans, and African Americans. Journal of Applied Social Psychology, 28, 2007-2024. Luhtanen, R., & Crocker, J. (1992). A collective self-esteem scale: Selfevaluation of one's social identity. Personality and Social Psychology Bulletin, 18, 302-318. Maccoby, E+ E., & Jacklin, C. N. (1974). The psychology of sex differences. Stanford, CA: Stanford University Press. *Mandina, G. H. (1979). Self-concept and reading achievement in schools of varying ethnic compositions (Doctoral dissertation, East Texas State University, 1979). Dissertation Abstracts International, 40(7-A), 3666. Markus, H., & Kitayama, S. (1991). Culture and the self: Implications for cognition, emotion, and motivation. Psychological Review, 98, 224253. Marsh, H. W. (1986). Global self-esteem: Its relation to specific facets of self-concept and their importance. Journal of Personality and Social Psychology, 51, 1224-1236. Marsh, H. W. (1987). The big-fish-little-pond effect on academic selfconcept. Journal of Educational Psychology, 79, 280-295. Marsh, H. W. (1990). Influences of internal and external frames of references in the formation of math and English self-concepts. Journal of Educational Psychology, 82, 107-116. Marsh, H. W. (1993). Relations between global and specific domains of self: The importance of individual importance, certainty, and ideals. Journal of Personality & Social Psychology, 65, 975-992. Marsh, H. W., & Richards, G. E. (1988). Tennessee Self-Concept Scale: Reliability, internal consistency, structure, and construct validity. Journal of Personality and Social Psychology, 55, 612-624. *Marsh, L. K. (1974). Self-esteem, achievement responsibility, and reading achievement of lower-class Black, White, and Hispanic seventh-grade boys (Doctoral dissertation, New York University, 1974). Dissertation Abstracts International, 35(10-A), 6514-6515. *Martinez, R., & Dukes, R. L. (1987). Race, gender, and self-esteem among youth. Hispanic Journal of Behavioral Sciences, 9, 427-443. Maslow, A. H. (1970). Motivation and personality. New York: Harper. May, J. W. (1981). Effects of age on color preference for Black and White infants and young children. Perceptual and Motor Skills, 52, 255-261. Mboya, M. (1988). A comparative analysis of self-esteem and academic achievement of Black and White high school students. School Psychology International, 9, 197-201. McAdoo, H. P. (1977). The development of self-concept and race attitudes of young Black children over time. In W. Cross (Ed.), Empirical research in Black psychology 111 (pp. 46-82). Washington, DC: National Institute of Education. McCarthy, J. D., & Hoge, D. R. (1982). Analysis of age effects in longitudinal studies of adolescent self-esteem. Developmental Psychology, 18, 372-379. McCarthy, J. D., & Yancey, W. L. (1971). Uncle Tom and Charlie: Metaphysical pathos in the study of racism and personal disorganization. American Journal of Sociology, 7, 648-672. *McDonald, R. L., & Gynther, M. D. (1965). Relationship of self and ideal-self descriptions with sex, race, and class in southern adolescents. Journal of Personality and Social Psychology, 1, 85-88. McGuire, W. J. (1984). Search for the self: Going beyond self-esteem and the reactive self. In R. A. Zurcher, J. Aronoff, & A. I. Robin (Eds.), Personality and the prediction of behavior (pp. 73-120). Orlando, FL: Academic Press. McGuire, W. J., & McGuire, C. V. (1988). Content and process in the experience of self. Advances in Experimental Social Psychology, 21, 97-144. McMillan, M+ (1988). The doll test studies--From cabbage patch to self-concept. Journal of Black Psychology, 14, 69-72.

RACIAL COMPARISONS OF SELF-ESTEEM *McNair, D., & Brown, D. (1983). Predicting the occupational aspirations, occupational expectations, and career maturity of Black and White male and female tenth graders. Vocational Guidance Quarterly, 32, 29-36. McRae, J. A. (1991). Rasch measurement and differences between women and men in self-esteem. Social Science Research, 20, 421-436. Mead, G. H. (1934). Mind, self, and society. Chicago: University of Chicago Press. Microsoft. (1997). Excel 97 [Computer software]. Redmond, WA: Author. Miller, N., & Pollock, V. E. (1994). Meta-analysis and some sciencecompromising problems of social psychology. In W. R. Shadish and S. Fuller (Eds.), The social psychology of science (pp. 230-261. New York: Guilford Press. *Morrison, K. E. (1974). An examination of self-concept as it relates to the selected school behaviors of Puerto Rican, Black and White senior high school students in Camden, New Jersey (Doctoral dissertation, Rutgers State University, 1974). Dissertation Abstracts International 36(4-A), 2108-2109. *Moses, E. G., Zirkel, P. A., & Greene, J. F. (1973). Measuring the self-concept of minority group pupils. Journal of Negro Education, 93-98. Mullen, B. (1991). Group composition, salience, and cognitive representations: The phenomenology of being in a group. Journal of Experimental Social Psychology, 27, 297-323. Mullis, A. K., Mullis, R. L., & Normadin, D. (1992). Cross-sectional and longitudinal comparison of adolescent self-esteem. Adolescence, 27, 51-61. Munford, M. B. (1994). Relationship of gender, self-esteem, social class, and racial identity to depression in Blacks. Journal of Black Psychology, 20, 151-174. Murphy, J., Alpert, B. S., Moes, D., & Somes, G. (1986). Race and cardiovascular reactivity: A neglected relationship. Hypertension, 8, 1075-1083. *Oanh, N. T., & Michael, W. B. (1977). The predictive validity of each of ten measures of self-concept relative to teachers' ratings of achievement in mathematics and reading of Vietnamese children and of those from five other ethnic groups. Educational and Psychological Measurement, 37, 1005-1016. O'Brien, E. J. (1991). Sex differences in components of self-esteem. Psychological Reports, 68, 241-242. *O'Callaghan, K. W., & Bryant, C. (1990). Non-cognitive variables: A key to Black American success at a military academy. Journal of College Student Development, 31, 121-126. *O'Connor, J. K. (1977). Perception of self, ideal self and teacher feelings in elementary and middle school children (Doctoral dissertation, Florida State University, 1977). Dissertation Abstracts International 38(4-A), 1899. Ogbu, J. U. (1994). From cultural differences to differences in cultural frames of reference. In P. M. Greenfield & R. R. Cocking (Eds.), Cross-cultural roots of minority child development (pp. 365-391). Hillsdale, NJ: Erlbaum. O'Malley, P., & Bachman, J. (1983). Self-esteem: Change and stability between ages 13 and 23. Developmental Psychology, 19, 257-268. *Osborne, J. W. (1995). Academics, self-esteem, and race: A look at underlying assumptions of the disidentification hypothesis. Personality and Social Psychology Bulletin, 21, 449-455. *Osborne, W. L., & LeGette, H. R. (1982). Sex, race, grade level and social class differences in self-concept. Measurement and Evaluation in Guidance, 14, 195-201. Overton, R. C. (1998). A comparison of fixed-effects and mixed (randomeffects) models for meta-analysis tests of moderator variable effects. Psychological Methods, 3, 354-379. Oyserman, D., & Markus, H. R. (1993). The sociocultural self. In J. Suls (Ed.), Psychological perspectives on the self (Vol. 4, pp. 187-220). Hillsdale, NJ: Erlbaum.

47

Pallas, A. M., Entwisle, D. R., & Alexander, K. L. (1990). Social structure and the development of self-esteem in young children. Social Psychology Quarterly, 53, 302-315. Parham, T. A., & Helms, J. E. (1985). Attitudes of racial identity and self-esteem of Black students: An exploratory investigation. Journal of College Student Personnel 26, 143-147. *Pascarella, E. T., Smart, J. C., Ethington, C. A., & Nettles, M. T. (1987). The influence of college on self-concept: A consideration of race and gender differences. American Educational Research Journal 24, 49-77. Paul, M. J., & Fischer, J. L. (1980). Correlates of self-concept among Black early adolescents. Journal of Y~guthand Adolescence, 9, 163-173. *Petersen, B., & Ramirez, M. (1971). Real-ideal self-discrepancy in Negro and Mexican American children. Psychology, 8, 22-26. *Phillips, D. A., & Zigler, E. (1980). Children's self-image disparity: Effects of age, socioeconomic status, ethnicity, and gender. Journal of Personality and Social Psychology, 39, 689-700. Phinney, J. S. (1988, June). The development of ethnic identity in adolescents. Paper presented at the Utah State University Workshop on Identity Formation: Theoretical and Empirical Issues, Logan, UT. Phinney, J. S. (1989). Stages of ethnic identity in minority group adolescents. Journal of Early Adolescence, 9, 34-49. Phinney, J. S. (1990). Ethnic identity in adolescents and adults: Review of research. Psychological Bulletin, 108, 499-514. Phinney, J. S. (1996). When we talk about American ethnic groups, what do we mean? American Psychologist, 51, 918-927. Phinney, J. S., & Alipuria, L. L. (1990). Ethnic identity in college students from four ethnic groups. Journal of Adolescence, 13, 171-183. *Phinney, J. S., Cantu, C. L., & Kurtz, D. A. (1997). Ethnic and American identity as predictors of self-esteem among African American, Latino, and White adolescents. Journal of Youth and Adolescence, 26, 165-185. Phinney, J. S., & Chavira, V. (1992). Ethnic identity and self-esteem: An exploratory longitudinal study. Journal of Adolescence, 15, 271-281. *Picou, J. S., Cosby, A. G., Curry, E. W., & Wells, R. H. (1977). Race and the formation of academic self-concept: A causal analysis. Southern Journal of Educational Research, 57-70. Piers, P. V. (1984). Piers-Harris Children's Self-Concept Scale: Revised manual. Los Angeles: Western Psychological Services. Pliner, P., Chaiken, S., & Flett, G. (1990). Gender differences in concern with body weight and physical appearance over the life span. Personality and Social Psychology Bulletin, 16, 263-273. Porter, J. (1971). Black child, White child. Cambridge, MA: Harvard University Press. Porter, J., & Washington, R. (1979). Black identity and self-esteem: A review of studies of Black self-concept, 1968-78. American Review of Sociology, 5, 53-74. Porter, J., & Washington, R. (1993). Minority identity and self-esteem. Annual Review of Sociology, 19, 139-161. Powell, G. J. (1985). Self-concepts among Afro-American students in racially isolated minority schools: Some regional differences. Journal of the American Academy of Child Psychiatry, 24, 142-149. *Powell, G. J., & Fuller, M. (1973). Black Monday's children: A study of the effects of school desegregation in self-concepts of Southern children. New York: Appleton-Century-Crofts. Powell-Hopson, D., & Hopson, D. S. (1988). Implications of doll color preferences among Black preschool children and White preschool children. Journal of Black Psychology, 14, 57-63. *Pratt, L. W. (1992). The effects of contingency contracting on race, academic achievement, classroom behavior, and self-esteem of low performing male elementary students (Doctoral dissertation, University of Florida, 1992). Dissertation Abstracts International 53(10-A), 3441 3442. Pyant, C. T., & Yanico, B. J. (1991). Relationship of racial identity and gender-role attitudes to Black women's psychological well-being. Journal of Counseling Psychology, 38, 315-322.

48

GRAY-LITTLE AND HAFDAHL

*Rich, P. M. (1973). Self-structure and social structure: A study of selfidentification and self-evaluation among Black and White college students (Doctoral dissertation, Purdue University, 1973). Dissertation Abstracts International, 34(6-A), 3544. *Richman, C. L., Clark, M. L., & Brown, K. F. (1985). General and specific self-esteem in late adolescent students: Race × gender × SES effects. Adolescence, 20, 555-566. *Riedel, J. S. (1980). Self-esteem, achievement scores and IQ scores among students of three ethnic groups in grades seven and eight (Doctoral dissertation, Northern Illinois University, 1980). Dissertation Abstracts International, 41(3-A), 996. *Rio, A. T. (1979). Defensiveness, self criticism and self concept in a sample of Black, Mexican, and White American adolescents (Doctoral dissertation, Michigan State University, 1979). Dissertation Abstracts International, 40(9-A), 4971. *Rodriguez, A. (1972). The relationship of self-concept and motor ability in certain selected Negro and Caucasian tenth grade girls (Doctoral dissertation, University of Alabama, 1972). Dissertation Abstracts International, 33(9-A), 4923. Rosenberg, M. (1965). Society and adolescent self-image. Princeton, NJ: Princeton University Press. Rosenberg, M. (1979). Conceiving the self. New York: Basic Books. Rosenberg, M. (1981). The self-concept: Social product or social force? In M. Rosenberg & R. Turner (Eds.), Social psychology: Sociological perspectives (pp. 593-624). New York: Basic Books, Rosenberg, M. (1986). Self-concept from middle childhood through adolescence. In J. Suls & A. G. Greenwald (Eds.), Psychological perspectives on the self(Vol. 3, pp. 107-136). Hillsdale, NJ: Erlbaum. Rosenberg, M., & Pearlin, L. I. (1978). Social class and self-esteem among children and adults. American Journal of Sociology, 84, 53-77. Rosenberg, M., Schooler, C., Schoenbach, C., & Rosenberg, F. (1995). Global self-esteem and specific self-esteem: Different concepts, different outcomes. American Sociological Review, 60, 141-156, Rosenberg, M., & Simmons, R. G. (1972). Black and White self-esteem: The urban school child. Washington, DC: American Sociological Association. Rosenthal, R. (1994). On being one's own case study: Experimenter effects in behavioral research--30 years later. In W. R. Shadish & S. Fuller (Eds.), The social psychology of science (pp. 214-229). New York: Guilford Press. Rotberam-Borus, M. J. (1989). Ethnic differences in adolescents' identity status and associated behavior problems. Journal of Adolescence, 12, 361-374. Rotheram-Borus, M. J. (1990). Adolescents' reference-group choices, selfesteem, and adjustment. Journal of Personality and Social Psychology, 59, 1075-1081. Rowley, S. J., Sellers, R. M., Chavous, T. M., & Smith, M. A. (1998). The relationship between racial identity and self-esteem in African American college and high school students. Journal of Personality and Social Psychology, 74, 715-724. *Rust, J. O., & McCraw, A. (1984). Influences of masculinity-femininity on adolescent self-esteem and peer acceptance. Adolescence, 19, 359366. Sattler, J. (1970) Racial "experimenter effects" in experimentation, testing, interviewing, and psychotherapy. Psychological Bulletin, 73, 137-160. *Savage, V. T. (1981). Expectancy, reinforcement-value, and demographics as predictors of situation-specific and general self-esteem (Doctoral dissertation, University of Illinois at Urbana-Champaign, 1981 ). Dissertation Abstracts International, 42(9-B), 3833. Schofteld, J. W., & Sagar, H. A. (1977). Peer interaction patterns in an integrated middle school. Sociometry, 40, 130-138. *Senior, A. D. (1989). Understanding differences in Black and White students' self-concept of ability: A examination of attitudes and achievement across several domains of self-concept (Doctoral dissertation,

University of Michigan, 1989). Dissertation Abstracts International, 51(1-B), 458. Shadish, W. R., & Haddock, C. K. (1994). Combining effect sizes. In H. Cooper & L. V. Hedges (Eds.), Handbook of research synthesis (pp. 285-299). New York: Russell Sage Foundation. Shrauger, J. S,, & Schoeneman, T. J. (1979). Symbolic interactionist view of self-concept: Through the looking glass darkly. Psychological Bulletin, 86, 549-573. Silverman, I., & Shaw, M. E. (1973). Effects of sudden mass school desegregation on interracial interaction and attitudes in one Southern city. Journal of Social Issues, 29, 133-142. Simmons, R. G., & Blyth, D. A. (1987). Moving into adolescence: The impact of pubertal change and school context. New York: Aldine de Grnyter. Singleton, L. C., & Asher, S. R. (1979). Racial integration and children's peer preferences: An investigation of developmental and cohort differences. Child Development, 50, 936-941. Skaalvik, E. M. (1986). Sex differences in global self-esteem: A research review. Scandinavian Journal of Educational Research, 30, 167-179. Skaalvik, E. M., & Hagtvet, K. H. (1990). Academic achievement and self-concept: An analysis of causal predominance in a developmental perspective. Journal of Personality and Social Psychology, 58, 292-307. Smith, E. J. (1991). Ethnic identity development: Toward the development of a theory within the context of majority/minority status. Journal of Counseling Psychology, 70, 181-188. Spencer, M. B. (1984). Black children's race awareness, racial attitudes, and self-concept: A reinterpretation. Journal of Child Psychology and Psychiatry, 25, 433-441. *Spigner, L. B. (1986). Changes in mental age, self-concept, and creative thinking in ethnically different 3- and 4-year-old Head Start students (Doctoral dissertation, Texas Woman's University, 1985). Dissertation Abstracts International, 46(1 l-A), 3245. *Spivey, W. L. (1976). A study of the self-concept and achievement motivation of Black versus White high school male achievers (Doctoral dissertation, California School of Professional Psychology, 1976). Dissertation Abstracts International, 36(9-B), 4711. SPSS. (1997). SPSS for Windows (Version 8.0) [Computer software]. Chicago: Author. Steele, C. M. (1992, April). Race and the schooling of Black Americans. Atlantic Monthly, 269(4), 68-78. Steele, C. M. (1997). A threat in the air: How stereotypes shape intellectual identity and performance. American Psychologist, 52, 613-629. *Stephan, W. G., & Rosenfield, D. (1978). Effects of desegregation on race relations and self-esteem. Journal of Educational Psychology, 70, 670679. *Stephan, W. G., & Rosenfield, D. (1979). Black self-rejection: Another look. Journal of Educational Psychology, 71, 708-716. Stewart, A. J., & Healy, J. M. (1989). Linking individual development and social change. American Psychologist, 44, 30-42. St. John, N. H. (1975). School desegregation outcomes for children. New York: Wiley. *Strang, W. J. (1972). The self-concepts of children in elementary school with differing proportions of Negro and White students (Doctoral dissertation, University of Alabama, 1972). Dissertation Abstracts International, 33(10-A), 5567. Stryker, S., & Serpe, R. T. (1994). Identity salience and psychological centrality: Equivalent, overlapping, or complementary concepts? Social Psychology Quarterly, 57, 16-35. Swann, W. B., Jr. (1996). Self-traps: The elusive quest for higher selfesteem. New York: Freeman. Tajfel, H., & Turner, J. C. (1986). The social identity of intergroup behavior. In S. Worchel & W. G. Austin (Eds.), Psychology of intergroup relations. Chicago: Nelson-Hall. *Tashakkori, A. (1993). Race, gender, and pre-adolescent self-structure: A

RACIAL COMPARISONS OF SELF-ESTEEM test of construct-specificity hypothesis. Personality and Individual Differences, 14, 591-598. *Tashakkori, A., & Thompson, V. D. (1991). Race differences in selfperception and locus of control during adolescence and early adulthood: Methodological implications. Genetic, Social, and General Psychology Monographs, 117, 135-152. *Taylor, K. E. (1991). The dilemma of difference: The relationship of the intellectual development, racial identity, and self-esteem of Black and White students to their tolerance for diversity (Doctoral dissertation, University of Maryland, 1990). Dissertation Abstracts International, 52(2-A), 443. Taylor, M. C., & Walsh, E. J. (1979). Explanations of Black self-esteem: Some empirical tests. Social Psychology Quarterly, 42, 242-253. *Thomas, L. C. (1978). A study of differences in self-concept, vocational preferences, and school achievement among Black and White high school students (Doctoral dissertation, Washington State University, 1978). Dissertation Abstracts International, 39(5-A), 2764. *Tracey, T. J., & Sedlacek, W. E. (1985). The relationship of noncognitive variables to academic success: A longitudinal comparison by race. Journal of College Student Personnel, 26, 405-410. Triandis, H. (1989). The self and social behavior in differing cultural contexts. Psychological Review, 96, 506-520. *Trowbridge, N. (1972). Self-concept and socioeconomic status in elementary school children, American Educational Research Journal, 9, 525537. *Tucker, B. S. (1977). The development of directionality of self-concept of low socioeconomic young children as it relates to grade, sex and race (Doctoral dissertation, Florida State University, 1977). Dissertation Abstracts International, 38(4-A), 2016-2017. *Turner, C. B., & Turner, B. F. (1982). Gender, race, social class, and self-evaluations among college students. Sociological Quarterly, 23, 491-507. Twenge, J. M. (1997, August). Changes in self-esteem over. time, 19651994: A meta-analysis. Poster presented at the 105th Annual Convention of the American Psychological Association, Chicago, IL. *Vail, A. M. (1973). I-E control, anxiety, frustrating, and self-concept of achieving and underachieving adolescent boys: A racial comparison (Doctoral dissertation, Fordham University, 1973). Dissertation Abstracts International, 34(8-B), 4062. Vevea, J., & Hedges, L. V. (1995). A general linear model for estimating effect size in the presence of publication bias. Psychometrika, 60, 419-435.

49

Wade, T. J., Thompson, V. D., Tashakkori, A., & Valente, E. (1989). A longitudinal analysis of sex by race differences in predictors of adolescent self-esteem. Personality and Individual Differences, 10, 717-729. Ward, S. H., & Braun, J. (1972). Self-esteem and racial preference in Black children. American Journal of Orthopsychiatry, 42, 644-647. *Washington, C. E. (1978). Relationship between self-concept, socioeconomic status, attitude toward school, and level of achievement of Black and non-Black junior college freshman (Doctoral dissertation, University of Southern Mississippi, 1978). Dissertation Abstracts International, 39(9-A), 5256. *Wendiand, M. M. (1969). Self-concept in southern Negro and White adolescents as related to rural-urban residence (Doctoral dissertation, University of North Carolina, 1968). Dissertation Abstracts International, 29(07B), 2642. *White, C., & Burke, P. J. (1987). Ethnic role identity among Black and White college students: An interactionist approach. Sociological Perspectives, 30, 330-331. *Widaman, K. F., & McMillan, D. L. (1992). Differences in adolescent self-concept as a function of academic level, ethnicity, and gender. Journal of Mental Retardation, 96, 387-404. *Wolf, T. M., Hunter, S. M., Webber, L. S., & Berenson, G. S. (1981). Self-concept, locus of control, goal blockage, and coronary-prone behavior pattern in children and adolescents: Bogalusa Heart Study. Journal of General Psychology, 105, 13-26. Wright, B. H. (1985). The effects of racial self-esteem on the personal self-esteem of Black youth. International Journal of Intercultural Relations, 9, 19-30. Wylie, R. C. (1979). The self-concept: Vol. 2. Theory and research on selected topics. Lincoln, NE: University of Nebraska Press. Wylie, R. C. (1989). Measures of self-concept. Lincoln, NE: University of Nebraska Press. *Yando, R., Seitz, V., & Zigler, E. (1979). Intellectual and personality characteristics of children: Social-class and ethnic-group differences. Hillsdale, NJ: Edbaum. *Zipper, B. (1972). The personality and self-concept characteristics of Negro and White delinquent and non-delinquent boys (Doctoral dissertation, University of North Carolina, 1972). Dissertation Abstracts International, 34(01B), 0431. *Zirkel, P. A., & Moses, E. G. (1971). Self-concept and ethnic group membership among public school students. American Educational Research Journal, 8, 253-264.

(Appendix follows)

50

GRAY-LITrLE AND HAFDAHL

Appendix

Moderator

and Sample-Size

Information

for the 261 Independent

Effect

Sizes

Included

Sample % Author(s) (publication year) [effect size]" A b b o t t (1981) A h r (1989) [1] A h r (1989) [2] A t o l a g b e (1975) [1] A t o l a g b e (1975) [2] A t o l a g b e (1975) [3[ A t o l a g b e (1975) [4] A t o l a g b e (1975) [5] A t o l a g b e (1975) [6] B a c h m a n (1970) [1] B a c h m a n (1970) [2] B a e h m a n (1970) [3] Barnes & Farrier (1985) Bennett & L u n d g r e n (1976) B e n s o n & Rentsch (1988) B l a c k b o u r n & B l a c k b o u r n (1987) Bowler, Ranch, & S c h w a r z e r (1986) Brack, Orr, & Ingersoll (1988) [1] Brack, Orr, & Ingersoll (1988) [2] Brack, On-, & Ingersoll (1988) [3] Bridgette (1970) [1] Bridgette (1970) [2] Bruch, Kunce, & E g e m a n (1972) [1] Bruch, Kunce, & E g e m a n (1972) [2] B u r b a c h & B r i d g e m a n (1976) [1] B u r b a c h & B r i d g e m a n (1976) [2] B u r g e r (1973) c [1] B u r g e r (1973) ~ [2] B u r g e r (1973) c [3] B u r g e r (1973) c [4] B u r g e r (1973) ¢ [5] B u r g e r (1973) c [6] B u r k e & Hoelter (1988) [1] B u r k e & Hoelter (1988) [2[ Carpenter & Busse (1969) [1] Carpenter & Busse (1969) [2] Carpenter & Busse (1969) [3] Carpenter & Busse (1969) [4] Castillo (1983) Cicirelli (1977) [1] Cicirelli (1977) [2] Cicirelli (1977) [3] V. D. C o l e m a n (1979) [1] V. D. C o l e m a n (1979) [2] Collins (1993) c Cornell, Delcourt, Goldberg, & B l a n d (1995) [1] Comell, Delcourt, Goldberg, & B l a n d (1995) [2] Cozart (1989) c Crain & B r a c k e n (1994) Crocker, Luhtanen, Blaine, & B r o a d n o x (1994) Dales & Keller (1972) Davids (1973) D a v i d s o n (1974) D e m o & Parker (1987) C. D. E d w a r d s (1972) [1] C. D. E d w a r d s (1972) [2] D. W. E d w a r d s (1974)

Outlet

A u t h o r race

Age

N

Male

Black

J D D D D D D D D J J J J J J J J J J J D D J J J J D D D D D D J J J J J J D J J J D D D J

---O O O O O O W W W -B W W W W W W W W ----------W W -----W W W -----

8.00 12.00 12.00 11.00 11.00 11.00 11.00 11.00 11.00 15.50 15.50 15.50 10.50 4.40 9.66 6.00 16.11 12.00 13.00 14.00 15.00 15.00 16.00 17.00 10.00 10.00 5.00 6.00 7.00 5.00 6.00 7.00 17.00 17.00 7.05 11.34 6.92 11.50 17.60 6.00 7.00 8.00 14.00 17.00 13.00 7.50

116 71 79 20 20 20 20 20 20 2,239 2,248 2,306 503 84 547 90 211 215 228 232 135 117 45 39 137 137 50 50 50 50 50 50 395 322 20 20 20 20 148 134 91 120 153 89 19,908 a 264

47 --0 0 0 100 100 100 100 100 100 45 100 60 52 -45 45 45 0 100 100 100 0 100 0 0 0 1130 100 100 0 100 0 0 100 100 -48 48 48 0 0 50 --

76 39 44 50 50 50 50 50 50 3 4 6 36 50 61 36 43 21 20 17 39 33 33 44 64 58 50 50 50 50 50 50 29 30 50 50 50 50 37 48 59 52 65 4 15 36

J

--

7.50

588

47

D J J

--W

11.60 13.50 19.00

112 2,095 187

J J D J D D J

-W -W --W

15.00 4.20 17.00 20.00 10.00 10.00 13.00

546 42 240 298 280 357 208

in Major

Analyses

School %

Instrument

SES

Black

White

Name b

St

Do

L

L M H L M H M M X L L

-27 27 ------90 60 90 ---

-73 73 ------90 90 90 ---

--

S

G

--

U

G

--

U

G

C C C C C C ---D --

S S S S S S U U U S S

G G G G G G G G G G G

--

--

M . M M M L L L L

35

--

G

S

G

65

--

24 24 24 100 100 ---

73 73 73 100 100 ---

I I I C C K K

S U S S S S S S S

G G G G G G G G G

--

--

--

C

S

G

--

--

--

C

S

G

L L L L L L X X L L L L M

-------------50

C C C C C C -------I

S S S S S S U U U U U U S

G G G G G G A A G G G G G

--

-------------50

H

S

G

--

50

50

H

S

G

--

50

50

H

S

G

--

--

--

L

S

G

--

--

--

L

S

G

---

.

.

.

. .

. .

. .

. .

U S

G A

30

.

.

.

.

S

A

-47 53

48 10 49

M . X

.

.

.

S S S

G G G

100 50 50 48 0 100 100

51 43 50 50 38 37 75

50

G

--

--

I

-57 -75 77

--

S

G

--

U

G

S S

G G

--

-43 -25 23

D

S

G

--

23

77

D

S

G

--

80

20

--

U

G

L --

M M

40

L L

RACIAL COMPARISONS OF SELF-ESTEEM Appendix

51

(continued)

Sample % Author(s) (publication year) [effect size] a

Outlet

A u t h o r race

Entwisle, Alexander, Pallas, & C a d i g a n (1987) [1] Entwisle, Alexander, Pallas, & C a d i g a n (1987) [2] Franklin (1985) Fu, Hinkle, & K o r s l u n d (1983) Gerken, Allen, & Snider (1984) Gray-Little & A p p e l b a u m (1979) [1] Gray-Little & A p p e l b a u m (1979) [2] G r a y - L i n l e & A p p e l b a u m (1979) [3] Gray-Little & A p p e l b a u m (1979) [4] G r e g o r y (1977) [1] G r e g o r y (1977) [2] G r e g o r y (1977) [3] G r e g o r y (1977) [4] G r e g o r y (1977) [5] G r e g o r y (1977) [6] G r e g o r y (1977) [7] G r e g o r y (1977) [8] G r e g o r y (1977) [9] G r e g o r y (1977) [10] G r e g o r y (1977) [11] G r e g o r y (1977) [12] Griffith (1985) [1] Griffith (1985) [2] Griffith (1985) [3] Griffith (1985) [4] Griffith (1985) [5] Griffith (1985) [6] Grossman (198l) H a r e (1980) [1] H a r e (1980) [2] Harris & Stokes (1978) H a r t n a g e l (1970) [1] H a r t n a g e l (1970) [2] Healey & DeBlassie (1974) H e n d e r s o n (1977) [1] H e n d e r s o n (1977) [2] H e n d e r s o n (1977) [3] H e n d e r s o n (1977) [4] H e n d e r s o n (1977) [5] H e n d e r s o n (1977) [6] H e n d e r s o n (1977) [7] H e n d e r s o n (1977) I8] H e r n a n d e z (1984) Hines & B e r g - C r o s s (1981) Hoelter (1982) [1] Hoelter (1982) [2] H o l a d a y , Callahan, Fabre, & Hall (1996) H o u s e (1993) [1] H o u s e (1993) [21 H o u s e l y et al. (1987) [1] H o u s e l y et al. (1987) [2] H u l b a r y (1975) J. G. H u n t & H u n t (1977) [1] J. G. H u n t & H u n t (1977) [2] J. P. H u n t (1991) H u r l e y (1973) c J a m e s o n (1973) [1] J a m e s o n (1973) [2] C. H. Jones (1973) [1] C. H. Jones (1973) [2] C. H. Jones (1973) [3] C. H. Jones (1973) [4]

J

School %

Age

N

W

6.50

341

0

52

.

.

.

J

W

6.50

332

100

56

.

.

D J J J J J J D D D D D D D D D D D D D D D D D D D J J J J J J D D D D D D D D D J J J J J J J J J J J D D D D D D D D

--W B B B B ------------------W B B ---W --------H B W W -----------------

19.00 10.00 8.87 12.31 15.33 15.33 12.31 13.00 13.00 13.00 16.00 16.00 16.00 13.00 13.00 13.00 16.00 16.00 16.00 15.50 15.50 15.50 15.50 15.50 15.50 13.50 10.50 10.50 -15.50 15.50 14.00 10.50 10.50 10.50 10.50 10.50 10.50 10.50 10.50 7.50 12.50 --14.97 18.20 18.20 15.00 15.00 16.50 13.00 16.00 12.00 16.00 13.00 9.00 9.00 10.00 11.00 9.00

84 1,372 238 188 159 165 209 903 791 842 567 614 1,020 679 710 754 353 441 733 366 382 1,333 2,922 3,386 5,114d 341 198 315 225 74 201 465 42 42 47 52 39 41 42 47 89 180 876 684 229 115 76 34 56 186 249 179 296 182 184 208 170 216 225 246

0 0

50 49 36 38 47 36 42 68 30 23 54 22 8 69 28 20 58 28 8 16 12 11 4 4 22 6 40 7 54 58 35 9 50 57 28 52 54 41 52 49 21 48 42 42 64 71 55 53 50 47 48 52 22 43 50 50 21 18 12 14

Male

-50 51 56 58 0 0 0 0 0 0 100 100 100 100 100 100 ------51 --100 100 100 52 0 0 0 0 100 100 100 100 48 47 0 100 -0 100 0 0 45 100 100 -100 0 100 0 0 0 100

Black

SES

Black

M .

White

St

Do

.

S

A

.

.

S

A

S

G

.

.

-.

Instrument

--

Name b

--

S

G

--

--

--

G

S

G

X X X X L M H L M H L M H L M H . . . . . . M L M

20 20 20 20 -------------

80 80 80 80 -------------

L L C C J J J J J J J J J J J J

4 23 23

65 77 77

I ---

S S S S S S S S S S S S S S S S U U U U U U S U U

G G G G G G G G G G G G G G G G G G G G G G G G G

--

--

--

L L

41 41

59 59

--

6

--

--

--

. . . . . .

. . . . . .

. . . . . .

I

S

G

---

U U

O O

70

L

S

G

--

G

S

G

--

--

G

S

G

--

--

--

G

S

G

--

--

--

G

S

G G

--

--

--

G

S

--

--

--

G

S

G

--

--

--

G

S

G

--

--

--

G

S

G

M

-40 42 42 100

-60 58 58 --

G

S

G

L

S

G

--

U

G

---

U S U U

G G A A

---

-X . .

. .

. .

. .

--

20

80

I

S

• G

--

20

80

I

S

G

--

--

--

I

S

G

X X L .

----

----

I I --

S S S U

G G G G G

.

.

.

--

--

--

C

S

--

--

--

C

S

G

--

16

64

C

S

G

--

16

64

C

S

G

--

16

64

C

S

G

--

16

64

C

S

G

(Appendix

continues)

52 Appendix

GRAY-LITTLE AND HAFDAHL

(continued) Sample %

Instrument

Outlet

Author race

Age

N

Male

Black

SES

Black

White

N a m eb

St

Do

--B -W W

10.120 11.00 15.70 5.29 17.00 16.00

203 215 98 42 20,029 d 35,099 d

100 100 0 52 46 50

12 16 49 71 15 11

--M . X L

16 16 ----

64 64 -. ---

C C --

&

D D J D J J

---

S S S U U U

G G G O G A

&

J

W

16.00

75,614 d

50

7

M

--

--

--

U

A

&

J

W

16.00

34,271 d

50

5

H

--

--

--

U

A

&

J

W

10.00

44,562 a

51

18

L

--

--

--

U

A

&

J

W

10.00

76,216 a

51

8

M

--

--

--

U

A

&

J

W

10.00

35,639 d

51

5

H

--

--

--

U

A

&

J

W

13.00

86,401 d

51

9

M

--

--

--

U

A

&

J

W

13.00

41,795 d

51

7

H

--

--

--

U

A

J D D D D D J J J J D D D D D D D D D D D D D J J J D D J J J J D J J J J D D D J J

W -H H H H B B -W ------------------H H W B --O ------W

7.00 9.50 11.00 11.00 11.00 11.00 17.00 17.00 17.00 12.93 9.00 10.00 12.00 14.00 15.00 16.00 9.00 10.00 1Z00 14.00 15.00 16.00 21.00 6.00 6.00 6.00 12.00 12.00 14.45 14.45 17.50 15.00 15.90 10.50 11.00 21.00 18.00 11.00 10.00 9.00 13.00 13.00

72 603 131 158 112 151 128 138 35,563 d 80 129 120 115 170 123 126 110 124 108 161 125 119 362 144 96 96 500 202 8,009 8,106 472 259 244 80 48 60 62 103 76 70 12,000 a 367

56 51 0 0 100 100 0 100 -100 0 0 0 0 0 0 100 100 100 100 1130 100 26 50 0 100 46 100 --44 48 43 50 -78 79 49 51 53 49 43

38 25 76 24 70 28 39 33 55 50 47 46 60 52 61 52 47 55 47 45 42 50 50 50 50 50 59 50 8 8 55 36 43 50 38 50 52 24 21 21 10 18

-X M M M M --. -L L L L L L L L L L L L . ---. L M M -X -L ------. X

25 20 ----34 34

60 80 ----65 65 . 100 16 16 16 16 16 16 16 16 16 16 16 16 . ---. -92 92 100 -25 --92 92 ---. 80

G -----L L

S S U U U U S S U S U U U U U U U U U U U U S S S S U S U U S S S S S S S S S S U S

G A A A A A G G A G A A A A A A A A A A A A G G G G A G G O O G G G G G G G G G G G

Author(s) (publication year) [effect size] a C. H. Jones (1973) [5] C. H. Jones (1973) [6] E. E. Jones (1979) Kearney (1973) Keith, Pottebaum, & Eberhart (1986) Kohr, Coldiron, Skiffington, Masters, Blust (1988) [1] Kohr, Coldiron, Skiffington, Masters, Blust (1988) [2] Kohr, Coldiron, Skiffington, Masters, Blust (1988) [3] Kohr, Coldiron, Skiffington, Masters, Blust (1988) [4] Kohr, Coldiron, Skiffington, Masters, Blust (1988) [5] Kohr, Coldiron, Skiffington, Masters, Blust (1988) [6] Kohr, Coldiron, Skiffington, Masters, Blust (1988) [7] Kohr, Coldiron, Skiftington, Masters, Blust (1988) [8] Kugle, Clements, & Powell (1983) Kuhn (1973) c L a r y e a (1972) [1] L a r y e a (1972) [2] L a r y e a (1972) [3] L a r y e a (1972) [4] L a w r e n c e & Brown (1976) [1] L a w r e n c e & B r o w n (1976) [2] Lay & Wakstein (1985) Lefebvre (1973) Leonards (1988) c [1] Leonards (1988) c [2] Leonards (1988) ~ [3] Leonards (1988) c [4] Leonards (1988) c [5] Leonards (1988) ~ [6] Leonards (1988) c [7] Leonards (1988) c [8] Leonards (1988) ¢ [9] Leonards (1988) c [10] Leonards (1988) ~ [11] Leonards (1988) c [12] L o g a n (1984) Long & Henderson (1968) Long & Henderson (1970) [1] Long & Henderson (1970) [2] Mandina (1979) L. K. Marsh (1974) Martinez & Dukes (1987) [1] Martinez & Dukes (1987) [2] McDonald & Gynther (1965) McNair & B r o w n (1983) Morrison (1974) Moses, Zirkel, & Greene (1973) Oanh & Michael (1977) O ' C a l l a g h a n & Bryant (1990) [1] O ' C a l l a g h a n & Bryant (1990) [2] O ' C o n n o r (1977) [1] O ' C o n n o r (1977) [2] O ' C o n n o r (1977) [3] J. W. Osborne (1995) W. L. Osborne & LeGette (1982)

School %

.

.

.

. 100 27 27 27 27 27 27 27 27 27 27 27 27

.

. ----

.

. -8 8 100 -60 --8 8 ----

.

. 20

L ------------B B B C ---L L C G --E E E G

53

RACIAL COMPARISONS OF SELF-ESTEEM Appendix

(continued)

Sample % Author(s) (publication year) [effect size] a Pascarella, Smart, Ethington, & Nettles (1987) Petersen & R a m i r e z (1971) Phillips & Zigler (1980) [1] Phillips & Zigler (1980) [2] Phillips & Zigler (1980) [3] Phillips & Zigler (1980) [4] Phinney, Cantu, & Kurtz (1997) [1] Phinney, Cantu, & Kurtz (1997) [2] Picou, C o s b y , Curry, & Wells (1977) Powell & Fuller (1973) [1] Powell & Fuller (1973) [2] Powell & Fuller (1973) [3] Powell & Fuller (1973) [4] Powell & Fuller (1973) [5] Powell & Fuller (1973) [6] Pratt (1992) [1] Pratt (1992) [2] Rich (1973) [1] R i c h (1973) [2] R i c h m a n , Clark, & B r o w n (1985) Riedel (1980) [1] Riedel (1980) [2] Riedel (1980) [3] Riedel (1980) [4] Rio (1979) R o d r i g u e z (1972) R u s t & M c C r a w (1984) S a v a g e (1981) Senior (1989) Spigner (1986) [1] Spigner (1986) [2] Spivey (1976) Stephan & Rosenfield (1978) Stephan & Rosenfield (1979) Strang (1972) T a s h a k k o r i (1993) T a s h a k k o r i & T h o m p s o n (1991) [1] T a s h a k k o r i & T h o m p s o n (1991) [2] K. E. T a y l o r (1991) c T h o m a s (1978) T r a c e y & S e d l a c e k (1985) T r o w b r i d g e (1972) T u c k e r (1977) [1] T u c k e r (1977) [2] T u c k e r (1977) [3] T u c k e r (1977) [4] T u c k e r (1977) [5] T u c k e r (1977) [6] T u c k e r (1977) [7] T u c k e r (1977) [8] T u r n e r & Turner (1982) [1] T u r n e r & T u r n e r (1982) [2] T u r n e r & T u r n e r (1982) [3] T u r n e r & T u r n e r (1982) [4] Vail (1973) [1] Vail (1973) [2] W a s h i n g t o n (1978) W e n d l a n d (1969) [1] W e n d l a n d (1969) [2] W e n d l a n d (1969) [3] W e n d l a n d (1969) [4] W e n d l a n d (1969) [5] W e n d l a n d (1969) [6]

Outlet

A u t h o r race

Age

N

Male

Black

J

W

19.00

4,597

46

17

J J J J J J J J J J J J J J D D D D J D D D D D D J D D D D D J J D J J J D D J J D D D D D D D D J J J J D D D D D D D D D

-W W W W W W

11.50 7.50 7.50 10.50 10.50 16.00 16.00 15.00 13.50 13.00 13.00 13.50 13.50 13.50 9.50 9.50 20.00 20.00 16.20 12.00 13.00 12.00 13.00 16.00 15.00 15.50 11.10 11.00 3.50 3.50 16.00 11.00 10.22 11.11 13.01 16.00 16.00 19.50 16.50 18.00 10.50 5.00 6.00 7.00 8.00 5.00 6.00 7.00 8.00 20.00 20.00 20.00 20.00 15.00 15.00 -13.78 13.78 13.78 13.78 14.04 14.05

90 20 20 20 20 146 151 179 259 298 336 357 176 294 36 36 331 448 195 82 67 81 103 252 176 172 193 2,605 10 17 40 837 243 964 637 9,593 d 10,325 d 199 120 1,789 3,789 50 50 50 50 50 50 50 50 51 71 37 18 59 60 250 42 89 104 123 93 78

46 50 50 50 50 0 100 100 39 40 43 46 49 51 100 100 0 100 52 0 0 100 100 -0 43 52 48 0 100 100 -54 49 49 -0 46 50 -51 0 0 0 0 100 100 100 100 0 0 100 100 100 100 38 0 0 0 0 100 100

26 50 50 50 50 79 77 45 36 54 46 58 53 22 50 50 41 30 75 35 31 32 26 40 50 13 31 7 40 65 50 37 21 53 47 25 27 24 50 12 18 50 50 50 50 50 50 50 50 27 21 57 39 49 50 58 24 71 55 54 70 42

B B B B B B ---------H -W -----W W -W O O --W ----------------W W W W W W

School % SES .

Black .

White

.

Instrument Name b

.

St

Do

U

0

L

--

--

--

U

G

L

--

--

--

U

G

M

--

--

--

U

G

L

--

--

--

U

G

M

--

--

--

U

G

--

--

--

I

S

G

--

--

--

I

S

G

U

A

.

.

.

.

L

--

--

L

S

G

X

--

--

L

S

G

X

--

--

L

S

G

L

--

--

L

S

G

X

--

--

L

S

G

X

--

--

L

S

G

--

--

--

G

S

G

--

--

--

G

S

G

L

--

--

--

S

G

L

--

--

--

S

G

--

75

25

I

S

G

--

7

86

C

S

G

--

7

86

C

S

G

--

7

86

C

S

G

--

7

86

C

S

G

--

37

48

L

S

G

--

25

75

--

U

G

--

--

--

G

S

G

X M L L

40 ----

60 ----

-----

U U U U

G O G G

--

L

S

G

U

G

--

.

--

.

.

.

--

15

55

I

S

G

--

--

--

C

S

G

U S S

G G G

. . .

. . .

. . .

. . .

--

7

91

I

S

G

--

--

--

L

S

G

S

A

.

.

.

.

--

--

--

C

S

G

L L L L L L L L L H L H M M .

------------25 25

F F F F F F F F ----K K

S S S S S S S S U U U U S S S

G G G G G G G G G G G G G G G

--

100

------------75 75 . 100

L

S

G

--

100

100

L

S

G

--

100

100

L

S

G

--

100

100

L

S

G

--

100

100

L

S

G

--

100

100

L

S

G

.

.

(Appendix

continues)

GRAY-LITtLE AND HAFDAHL

54 A p p e n d i x (continued)

Sample % Outlet

Author race

Age

N

Male

Black

SES

(1992) & Berenson

D D J J J

W W B W W

14.06 14.06 19.50 13.00 14.50

31 125 212 793 40

100 100 -63 30

23 37 34 24 40

---. L

& Berenson

J

W

16.50

50

32

58

& Berenson

J

W

15.50

52

33

& Berenson

J

W

12.50

54

& Berenson

J

W

17.50

& Berenson

J

W

& Berenson

J

& Berenson (1979) (1979) (1979) (1979) (1979) (1979) (1979) (1979)

Author(s) (publication year) [effect size] a Wendland (1969) [7] Wendland (1969) [8] White & Burke (1987) W i d a m a n & McMillan Wolf, Hunter, Webber, (1981) [11 Wolf, Hunter, Webber, (1981) [2] Wolf, Hunter, Webber, (1981) [3] Wolf, Hunter, Webber, (1981) [4] Wolf, Hunter, Webber, (1981) [5] Wolf, Hunter, Webber, (1981) [6] Wolf, Hunter, Webber, (1981) [7] Wolf, Hunter, Webber, (1981) [8] Yando, Seitz, & Zigler Yando, Seitz, & Zigler Yando, Seitz, & Zigler Yando, Seitz, & Zigler Yando, Seitz, & Zigler Yando, Seitz, & Zigler Yando, Seitz, & Zigler Yando, Seitz, & Zigler Zipper (1972) Zirkel & Moses (1971) Zirkel & Moses (1971) Zirkel & Moses (1971)

School %

[1] [2] [3]

[1] [2] [3] [4] [5] [6] [7] [8]

Black

Instrument

White

Name b

St

Do

100 100 -. . --

100 100 -. --

L L I G

S S S S S

G G G G G

L

--

--

G

S

G

56

L

--

--

G

S

G

37

37

L

--

--

G

S

G

38

37

76

L

--

--

G

S

G

13.50

36

50

44

L

--

--

G

S

G

W

11.50

42

52

38

L

--

--

G

S

G

J

W

10.50

36

53

44

L

--

--

G

S

G

J J J J J J J J D J J J

W W W W W W W W W W W W

8.30 8.30 8.30 8.50 8.50 8.50 8.50 8.50 14.00 10.50 10.50 10.50

48 80 48 48 80 48 80 80 82 20 30 30

50 50 50 50 50 50 50 50 100 50 50 50

50 50 50 50 50 50 50 50 49 50 33 67

L L M L L M M M L M M M

--------20 17 30 48

--------80 23 60 29

---A A A A --C C C

U U U S S S S U U S S S

G G G G G G G G G G G G

Note. Dash indicates information neither reported nor available from author(s). Outlet = publication outlet; D = dissertation, J = journal or book. Author race: B = Black; H = Hispanic; O = other; W = White. Age = average participant age for the (sub)sample; the two decimal places listed do not necessarily reflect precision of reported age in the primary study. N = total number of respondents included in the comparison for this effect size (does not reflect restricted weighting). Sample % = percentage in effect-size (sub)sample, where base rate is N. SES = socioeconomic status; L = low; M = middle; H --high; X = mixed or indeterminate, School % = percentage in school, where base rate is the size/population of the entire school(s) from which the respective racial subsample was drawn; for some studies, Black and White students were drawn from different schools. Instrument name: A = Brown Self-Concept Scale; B = Children's Self-Social Constructs Test; C = Coopersmith Self-Esteem Inventory; D = Lipsitt Self-Concept Scale; E = Lipsitt Self-Concept Scale--Discrepancy; F = Pictorial Self-Concept Scale; G = Piers-Harris Self-Concept Scale; H = Purdue Self-Concept Scale; I = Rosenberg Self-Esteem Scale; J = Self-Observation Scales--Social Acceptance; K = Semantic Differential; L = Tennessee Self-Concept Scale. St = instrument standardization; S = standardized; U = unstandardized. Do = instrument domain; A = specific-academic; G = general; O = other specific. a Bracketed arbitrary number to identify uniquely each within-study effect size. b Only standardized instruments used for three or more effect sizes are identified by name. c The listed publication year minus one (e.g., 1972 instead of 1973), which reflects the year the dissertation was completed, was used in analyses, d Weight restricted to 700 (see Footnote 5 and the ES weights paragraph under Statistical Methods in the Method section). R e c e i v e d J u l y 22, 1997 R e v i s i o n r e c e i v e d J u n e 22, 1999 A c c e p t e d J u l y 16, 1999 •