Using the Thorndike Model to Assess the Fairness of Cognitive Ability Tests for Personnel Selection
Abstract
This study evaluates cognitive ability tests (CATs) as predictors of job performance against the Thorndike (1971) model of fairness. Meta-analytic results indicate that CATs substantially misrepresent the relative qualifications between Blacks and Whites in the United States: CATs predict an average job performance difference between groups as three times larger than is actually the case. In practice, then, Blacks are disproportionally burdened by more false-negative selection errors, and this tendency increases markedly under higher CAT cutoffs. Thus, CATs work against proportionate representation of Blacks in the workplace. From an Employment Equity (E.E.) perspective, this is not justifiable because the pool of qualified Black candidates, relative to Whites, is considerably larger than is suggested by CAT scores.
Document Details
- Document Type
- Technical Report
- Publication Date
- May 01, 2001
- Accession Number
- ADA398473
Entities
People
- Greg A. Chung-yan
- Steven F. Cronshaw
Organizations
- University of Guelph