| Abstract: | In defending tests of cognitive abilities, knowledge, or skills (CAKS) from the skepticism of their “family members, friends, and neighbors” and aiding psychologists forced to defend tests from “myth and hearsay” in their own skeptical social networks (p. 215), Sackett, Borneman, and Connelly (see record 2008-05553-001) focused on evaluating validity coefficients, racial or gender group differences, and fair assessment research. In doing so, they concluded that CAKS tests generally yield valid and fair test scores for their intended purposes, but because the authors did not adequately attend to (a) research design issues (e.g., inclusion of independent or predictor variables [IPV] and dependent variables or criteria), (b) statistical assumptions underlying interpretation of their analyses (e.g., bivariate normality of distributions of test scores and criteria), and (c) conceptual concerns (e.g., whether racial categories should be used as explanatory constructs), alternative conclusions about CAKS test score validity and fairness are plausible. Although all of the foregoing areas of concern are germane to each of the assertions addressed by Sackett et al. (2008), the focus here is on Assertions 6 through 8 (p. 216; hereinafter called the fairness assertions [FA]) because making accurate inferences about fairness requires measurement experts to engage in a paradigmatic shift where sociodemographic groups (e.g., Blacks, Latinos/Latinas) are concerned, whereas, for the most part, addressing the other assertions merely requires a reminder of which standard psychometric principles have not been followed (American Educational Research Association [AERA], American Psychological Association [APA], & National Council on Measurement in Education [NCME], 1999). (PsycINFO Database Record (c) 2010 APA, all rights reserved) |
