Search Results

Title: Cognitive Skills and Racial Wage Inequality: Reply To Farkas And Vicknair
Resulting in 1 citation.
1. Maume, David J. Jr.
Cancio, A. Silvia
Evans, T. David
Cognitive Skills and Racial Wage Inequality: Reply To Farkas And Vicknair
American Sociological Review 61,4 (August 1996): 561-564.
Also: http://www.jstor.org/stable/2096393
Cohort(s): NLSY79
Publisher: American Sociological Association
Keyword(s): Armed Forces Qualifications Test (AFQT); Behavior; Cognitive Ability; Family Background; Racial Differences; School Quality; Test Scores/Test theory/IRT; Tests and Testing

Permission to reprint the abstract has not been received from the publisher.

Farkas and Vicknair (1996) create a composite measure of cognitive skills that is the average standardized score on tests of word knowledge, paragraph comprehension, arithmetic reasoning, and mathematical knowledge. These four tests are known as the "Armed Forces Qualifications Test" (AFQT), the scores the military uses to select recruits and for job assignments. Farkas and Vicknair claim to explain away the discrimination component of the racial gap in wages by controlling for cognitive skills. Yet they ignore the possibility that the AFQT, like many standardized tests, is class-biased. If these tests (especially those stressing a knowledge of vocabulary) also test for exposure to the values and experiences of the White middle class (Schiff and Lewontin 1986:33), then Blacks' scores on these tests will be systematically lower than scores for Whites. Indeed, Wigdor and Green (1991:179) reviewed studies of the AFQT and found larger Black/White differences on the AFQT than were found using direct measures of job performance. Thus, AFQT test scores exaggerate racial differences in skills and may in part be a proxy for race. If Blacks have lower scores on the AFQT and receive lower earnings because of employer discrimination, then inclusion of the AFQT score in a wage attainment model will weaken the impact of race on earnings. The AFQT score should be purged of its correlation with race so that the impact of cognitive skills, apart from race, can be examined. When the Department of Labor considered using military intelligence tests in the U.S. Employment Service, it commissioned a study that found "scientific grounds for the adjustment of minority scores so that able minority workers have the same chance of referral as able majority workers" (Hartigan and Wigdor 1989:7). To adjust AFQT scores received by minorities, Rodgers and Spriggs (1995:22) suggest regressing the scores on family background and school quality characteristics by race. The s lopes for Whites are then applied to values of the predictors for Blacks to generate a predicted AFQT score for Blacks. This method corrects for the devaluation of Blacks' scores by allowing the family background and school quality inputs to be evaluated in a "nondiscriminatory" manner. This method produces an instrumental variable that more closely approximates the job-relevant skills of African Americans. We replicated part of the Rodgers and Spriggs (1995) analysis using Farkas and present the results of regressing the AFQT composite score on family background and school quality variables. The data are for young males in the 1980 NLSY. Significant racial differences suggest that the AFQT test measures the abilities of Blacks and Whites differently.
Bibliography Citation
Maume, David J. Jr., A. Silvia Cancio and T. David Evans. "Cognitive Skills and Racial Wage Inequality: Reply To Farkas And Vicknair." American Sociological Review 61,4 (August 1996): 561-564.