An analytical comparison of conventional and tailored tests.

Abstract

Tailored tests were compared to 'best fixed' tests consisting of optimally chosen sets of items for all examinees. The results indicated that while the fixed tests provided very little information at the extremes of the ability distribution, the fixed tests provided as much or more information than the tailored tests in some regions of the ability distribution. The fixed tests provided average information values close to those for tailored tests. Therefore, if testing is used for selection rather than for classification, fixed tests can serve as competitors to tailored tests.