ITEM BIAS INDEXES BASED ON TOTAL TEST SCORE AND JOB-PERFORMANCE ESTIMATES OF ABILITY

Researchers (e.g., Ironson, 1982; Tenopyr, 1990) have suggested that i tem bias investigators equate subgroups on external criteria such as j ob performance rather than total test scores before considering subgro up passing rates on test items. In a study comparing these two approac hes to studies of item bias, we found little evidence of bias using to tal test score as the estimate of overall examinee ability, but nearly all items were biased in comparisons of white and African-American su bgroups on Numerical, Verbal, and Mechanical Reasoning tests and in ma le-female comparisons on a Mechanical Reasoning test when job performa nce was used to select ''equally able'' examinees. However, the use of job performance as the ability index is analogous to performance-base d approaches to test bias (Hartigan & Wigdor, 1989; Thorndike, 1971) a nd directly equivalent to the Darlington (1971) and Cole (1973) test b ias definition, the logical inconsistencies of which have been previou sly described (Hunter & Schmidt, 1976; Peterson & Novick, 1976). We co nclude that performance matching as a basis of forming ''equal ability '' groups is inappropriate.