Measurement Invariance of the American Identity Scale among White, Black, and Latino Adolescents

• In Event: 3-066 - “What’s Measurement Got to Do with It?” Testing Ethnic-Racial Equivalence of Identity-Related Measures

Integrative Statement

Adolescence is a developmental period when youth refine their sense of self and when identity formation is particularly salient (Erikson, 1968). As adolescents in the U.S. engage in this process, they recognize various social groups of which they may be a part, including their identity as a U.S. American. However, the meaning of one’s U.S. American identity may vary depending on one’s ethnic-racial (E-R) group membership. Indeed, the socio-historical context of the U.S. is such that some groups have been more easily integrated (e.g., White European groups) into U.S. mainstream culture than others (e.g., Black, Latino). To test research questions that compare American identity across E-R groups, however, measures must demonstrate psychometric equivalence. The current study examined measurement equivalence of the American Identity Scale (AIS), which assesses youths’ sense of belonging to the U.S. (Phinney & Devich-Navarro, 1997), across White, Black, and Latino adolescents.

Participants (N = 1,378; Mage = 16.16; SD = 1.12; Range = 13-19; 53% female; 49.6% White, 23.6% Black, 26.9% Latino) were recruited from a Southwestern high school and completed self-administered surveys. Most participants (96.2%) were U.S. born. Cronbach’s alpha for the AIS was identical (i.e., alpha = .92) for Whites, Blacks, and Latinos. Limited work examines E-R group invariance of the AIS. Therefore, we first explored the factor structure of the AIS with each E-R group using exploratory factor analysis (EFA). We then tested whether the factor structure, item loadings, item means, and latent factor means could be constrained to be equal across E-R groups using multigroup confirmatory factor analysis (CFA). We randomly split our sample in half and conducted our EFAs with the first half of the sample and used the second half of the sample to test our CFAs. Findings from the EFA suggested a one-factor structure for Black and Latino adolescents, and a two-factor structure for White adolescents (Table 1).

Given the similarity in factor structure for Black and Latino adolescents, a multi-group CFA was conducted to test measurement invariance of the AIS. White youth were not included in this model due to the differences in factor structure found in EFA and CFA. Results from sequentially constrained chi-square difference tests indicated that the factor loadings (i.e., metric invariance), intercepts (i.e., scalar invariance), and factors means (i.e., latent factor mean invariance) could be constrained to be equal among Black and Latino participants (Table 2).

Results suggest the AIS functions differently for White versus Black and Latino adolescents. Considering the socio-historical context of the U.S., it may be difficult to develop measures that equivalently assess American identity among members of historically marginalized (e.g., Black, Latino) and privileged (i.e., White European) groups. The current study provides empirical support for this notion and contributes to the literature by examining the psychometric properties of the AIS across E-R groups. Findings will be discussed with attention to the need for qualitative work to explore potential strategies for developing culturally informed measures of American identity that are psychometrically equivalent among members of diverse ethnic-racial groups.

Authors

• Stefanie Martinez-Fuentes, Arizona State University
  Presenting Author

• Adriana Umaña-Taylor, Harvard University
  Non-Presenting Author

• Elana R. McDermott, Harvard University
  Non-Presenting Author