Applied latent trait theory to an analysis of a 64-item multiple-choice skill test administered to 1,035 police recruits to illustrate how 2 shorter tests measuring the same attribute, but having different test characteristic curves, have different degrees of adverse impact. Specifically, ability differences between groups (195 Blacks, 373 Whites) were minimized by use of a 22-item wide-range test and exaggerated by use of a 22-item narrow-range test. In light of these data, claims such as those that performance tests have less adverse impact than paper-and-pencil tests are open to question; the differences attributed to the type of test are completely confounded with the differences in psychometric properties of the tests. The authors propose that the concept of adverse impact be redefined in terms of the degree to which test scores distort any underlying true subgroup differences in the attribute measured. (13 ref) (PsycINFO Database Record (c) 2006 APA, all rights reserved).
- latent trait analysis, distribution of scores on wide- vs narrow-range tests of same ability, Black vs White police recruits, implications for adverse impact ratios
ASJC Scopus subject areas
- Applied Psychology