Race and academic disidentification.

Summary of Findings In general, African American boys appear to be the onlygroup that experienced serious and significant disidentifica-tion with academics. No other group evidenced of signifi-cant disidentification. Hispanic girls manifested the stron-gest academic identification of any group studied. Thisdiscrepancy between two stigmatized groups raises someinteresting questions. Caveats Because these analyses used secondary data, much of thediscussion of identification is based on the inference that therelationship between self-esteem and academic outcomes isan indicator of identification with academics. Although thisis consistent with theory, in future studies a well-validatedmeasure of identification will be more critical, particularlyinstruments that can measure identification on an individual,rather than group, level (see Osbome, 1997; Voelkl, 1996).Other variables were not measured as reliably as theycould have been. In particular, self-esteem was measured ona shortened form of the Rosenberg (only 7 out of 10 itemsused). While still yielding internal reliabilities that areperfectly respectable and acceptable (.78-.82), this scale isnoted for impressive internal reliabilities (often ranging upto .91, as reported in Osborne, 1997). Interestingly, theprimary effect of decreased reliability is attenuated correla-tions, the statistic on which this article rests. Becauseinternal reliability did not change appreciably for any groupacross time, and because of the unique nature of ourhypotheses—looking for significant correlations to decreaseover time—this lower than normal reliability would only bea concern if we had observed null relationships across alltimepoints for a group.Overall, these caveats, although important to note, do notcall the major findings of the study into question. However,other issues need to be addressed. Some researchers havecriticized the Rosenberg Scale as being insensitive to theactual self-esteem of African Americans. Although thevalidity of this scale is open to debate, this sort of criticismdoesn't account for the fact that the correlations involvingAfrican Americans are the most dynamic of any group in thestudy. One would expect the opposite if insensitivity were areal concern. It is also likely that there are many confound-ing issues that occur during adolescence that this researchcannot account for. However, unless one can make acompelling argument that African American boys experi-ence different issues than Whites during adolescence asidefrom stigma and negative stereotypes, this criticism alsocannot account for the results observed, as most adolescentsremain significantly, if not strongly, identified with academ-ics.In sum, although this methodology has shortcomings italso affords certain advantages, such as an otherwise impos-sibly large sample that provides nationally representative,longitudinal data collected systematically and in a standard-ized way over a 4-year period, as well as data gathered frommultiple sources, such as transcripts, parents, and administra-tors. It would be impossible for anyone other than thegovernment to assemble this large, representative, andcomprehensive of a data set. The generalizability of thesefindings are a significant advantage to this study. Althoughthe results from analyses of secondary data are rarelydefinitive, they often provide important direction for futureresearch, and as such, serve a vital purpose in any field.