A Remedy for the Black-White Test-Score Disparity [Comment]

Abstract

Sackett, Schmitt, Ellingson, and Kabin (April 2001) analyzed the effectiveness of strategies for reducing the disparities in average scores on high-stakes tests of cognitive abilities (CATs) of (especially) African or Black and Latino and Latina Americans as compared with White Americans. They argued that decision makers in the domains of education, employment, and licensure and certification are becoming increasingly dependent on test scores as the primary criteria for making high-stakes decisions. Consequently, these two socioracial groups, as well as Native and Asian Americans (with respect to tests of verbal skills), who are already underrepresented in many selective educational institutions and professions, may disappear from them entirely if the disparities in test scores cannot be eliminated or rendered meaningless for making high-stakes decisions involving them. Sackett et al.'s proposed solutions to the problem are to either "dumb down" (i.e., remove cognitive content of) the tests or alter the testing process so that it appears to be fair to Black and Hispanic test takers, even, as if the authors' analysis suggested, it is not. In what seems to be an effort to prove that the test performance disparities between groups reflect actual irremediable cognitive deficiencies of the adversely affected test takers, Sackett et al. (2001) cited DeShon, Smith, Chan, and Schmitt's (1998) "unique study" (Sackett et al., 2001, p. 309) as disproving a "social relations and social context" (p. 309) argument, which they misattributed to me (Helms, 1992). I did not recommend that CAT items be modified to include social content. Most CATs already include such content. Instead, I discussed the absence of empirical evidence that CATs are culturally equivalent for African American test takers and proposed a strategy for quantifying the effects of racial and cultural variables on African, Latino and Latina, Asian, and Native American (ALANA) test takers' CAT scores. DeShon et al. allegedly collected the type of cultural data (e.g., racial identity attitudes) that could be used for trying the strategy but did not analyze it appropriately.