School Racial Composition and Biracial Adolescents' School Attachment

Abstract

Despite extensive research on multiracial youth in recent years, to date, no empirical studies have analyzed how racial context may affect biracial adolescents' sense of belonging in a social institution beyond families. In this study, we examine how the racial makeup of the student body affects self-identified biracial adolescents' school attachment. Using data from the National Longitudinal Study of Adolescent Health, we find that the proportions of white or black students in school significantly affect the school attachment of Hispanic/black, Asian/black, and American Indian/black biracial adolescents, but school racial composition in general has little influence on biracial adolescents with a partial-white identification (i.e., black/white, Hispanic/white, Asian/white, and American Indian/white). Our analyses also show that on average, students of most biracial groups display lower school attachment than their corresponding monoracial groups, but the differences from the monoracial groups with the lower school attachment are generally small. We discuss the implications of our findings for biracial adolescents' perceived racial boundaries and contemporary American race relations.

ACKNOWLEDGMENTS

We thank Brian Powell, Jason Beckfield, Brian Starks, Ralph McNeal Jr., Regina Werum, and anonymous reviewers for their helpful comments. We also thank Sarah Jacobson for her research assistance. This research was supported by grants from the University of Connecticut. The research uses data from Add Health, a program project designed by J. Richard Udry, Peter S. Bearman, and Kathleen Mullan Harris, and funded by a grant P01-HD31921 from the National Institute of Child Health and Human Development, with cooperative funding from 17 other agencies. Special acknowledgment is due Ronald R. Rindfuss and Barbara Entwisle for assistance in the original design. Persons interested in obtaining data files from Add Health should contact Add Health, Carolina Population Center, 123 W. Franklin Street, Chapel Hill, NC 27516-2524; e-mail: [email protected]

NOTES

Notes

¹ The field of multiracial studies has grown exponentially in the past decade. For a brief discussion on the social psychological consequences of multiracial self-identification, see Brown et al. (2006:415–16). For a review of outcome differences between biracial and monoracial adolescents (mostly based on parents' racial identifications), see Shih and Sanchez (2005). For discussions on the consequences of biracial individuals' fluid identities for racial politics, see the works by Brunsma and Rockquemore (e.g., Brunsma and Rockquemore 2001).

² Although our study focuses on the outcome profiles of biracial youth, this social psychological discussion can also be applied to the formation and maintenance of (multi)racial self-identifications. Relying on a similar social psychological argument, for example, Hitlin et al. (2006), examine the development and stability of multiracial self-identifications. They suggest that youth whose mothers have less education are more likely to switch between a monoracial and a multiracial identification than they are to adopt a consistent monoracial identity over time, whereas children of highly educated mothers are more likely to adopt a consistent multiracial identity (also see Doyle and Kao 2007a). Because having parents of distinct racial backgrounds is among the most important predictors for youth's multiracial self-identification, research (mostly about partial-white children) has also examined how interracial couples categorize their children's race (Brunsma 2005; Roth 2005). These studies suggest that high-SES parents are more likely to categorize their child as either multiracial or monoracial white as opposed to a monoracial nonwhite identity, and that the presence of racial minorities in the child's school or neighborhood increases the chances of parents categorizing their children as multiracial.

³ Add Health also allows a measure of adolescents' race based on their parents' racial identifications. We chose youth's racial self-identification in school over parent-based measures for two additional considerations. First, the parent-based measure is derived from one household member's (mostly mother's) identification of the two parents' races. Yet scholars have reported that parents are also inconsistent in identifying their children's or their own race (Udry et al. 2003; Brunsma 2005). This makes it difficult to determine whether parent-based race measures reflect adolescents' internal identity, expressed identity, or racial status. Second, Add Health does not collect biological parents' races if the parent does not live with the focal child. This requires researchers to exclude children who do not live in two-biological-parent households. Scholars have argued that this sample selection strategy is problematic (Harris and Sim 2002). In additional analyses, we assess the effects of school racial composition on white biracial, other biracial, and monoracial adolescents using 661 biracial adolescents identified from parent-based race. These analyses show consistent patterns with those reported here (available upon request). Unfortunately, the biracial sample from the in-home interviews is small and allows only limited differentiation of multiracial heterogeneity.

⁴ Because our coding of Hispanic-origin adolescents operates under the assumption that these students may think of Hispanic as their race, we use “racial identification” rather than “racial/ethnic identification” when referring to these students. Acknowledging that this phrasing may differ from that used in some studies on Latino individuals (e.g., Hitlin et al. 2007), we emphasize that the analytic distinction between “race” and “ethnicity” may not be as salient to individuals who identify themselves using these categories (Brown et al. 2006).

⁵ Ideally, we would prefer to have measures of how self-identified multiracial adolescents differ in their identifications with one or more of their associated monoracial groups. Unfortunately, this information is not available in the Add Health in-school data. Without the detailed information on individual self-identified multiracial students' affiliations to a monoracial group, in this paper, we can only rely on empirical reports from previous studies regarding how the general patterns of their racial affiliations may mediate the effects of school racial composition on their school attachment.

⁶ Of the 90,118 students, 4,597 non-Hispanic adolescents selected “other” or had a missing response, 1,195 adolescents selected more than two races, and 1,198 adolescents self-identified as Asian/Hispanic, American Indian/Hispanic, or American Indian/Asian. Of the remaining 83,128 students, 7,876 students had missing values in the dependent variable. A final group of 6,000 cases were also dropped from the analysis because of missing values in sampling weights. This leads to a final sample of 69,252 students. As for control variables, 24.4 percent of the students have missing values, mostly from GPA (11.1 percent) and parental education (12.5 percent). These students are more likely to be male, racial minority, and biracial, and they tend to have lower parental education. These students, however, attended similar schools and had comparable GPAs with those without missing values in the control variables. The non-randomness of missing cases in control variables precludes the use of listwise deletion and mean substitutions.

⁷ In supplementary analyses, we ran models using Erkut and Tracy's (2002) measure of school attachment, which included two additional Add Health items (“The teachers at this school treat students fairly” and “I feel safe in my school”; reliability score = .76). The results are consistent with those we present in the paper.

⁸ In Table 1, we also see a low proportion of American Indian female students and a high proportion of American Indian/black who are girls. Although most empirical research indicates no significant gender effect on adopting a multiracial self-identification (e.g., Brunsma 2005; Hitlin et al. 2006; Doyle and Kao 2007a), these studies did not look at American Indian/black youth. Thus, it may be interesting to explore the gender effect of identifying with both African Americans and American Indians.

⁹ We also estimated HLM models including Asian/Hispanic and American Indian/Hispanic students. The results suggest that the school attachment of Asian/Hispanic teens is not affected by the proportions of white, black, or Asian/Hispanic peers in school, whereas American Indian/Hispanic youth express stronger attachment to schools with higher proportions of black peers. We opt to not report these findings in our analyses given the focus of our paper on partial-white and partial-black biracial adolescents. It is also not clear if American Indian/Hispanic youth are expressing identification with two different groups or if they are non-American Indian Hispanics who think that “Native American” best describes the social category that they belong to.

¹⁰ We focus on the proportions of white and black peers for several considerations. One of our considerations is that the proportions of Asian or American Indian students are very small in most schools. This raises the question as whether the results from these analyses are statistical artifacts affected by a few influential cases, especially given the small numbers of cases in most biracial groups. This said, in ancillary analyses, we examined the effects of (1) the proportion of Hispanic peers on the school attachment of Hispanic/white adolescents, (2) the proportion of Asian peers on the school attachment of Asian/white adolescents, and (3) the proportion of American Indian peers on American Indian/white adolescents. The results suggest that the proportion of Hispanic peers in school has a significantly positive effect on Hispanic/white students, but the effects of the proportion of Asian or American Indian peers are statistically insignificant on Asian/white or American Indian/white adolescents.

¹¹ We conducted similar analyses for each of the racial groups we consider in this study. The results suggest that racial self-selection factors are negligible for monoracial, black/white, and American Indian/black biracial adolescents, but may affect the link between school racial composition and the school attachment of the other biracial groups. These analyses are only approximate checks because the treatment-effect model is not designed for multilevel analysis. To apply the model, we controlled for school clusters and conducted separate analyses for each racial group. School racial composition is used as the treatment, coded as 1 if the proportion of students from specific racial backgrounds (e.g., black peers, similar-race peers) exceeds .50. Student-level variables are included in the treatment equation to control for the confounding effects caused by non-racial factors. Some of the models estimated in these analyses do not converge because of flat numerical derivatives.