ABSTRACT
Measures of differential item functioning (DIF) can help test developers to ide_B.tify questions that may be unfair for members of sampled groups. DIF can, therefore, be an extremely useful tool for test developers. The use of DIF in test development, however, raises many questions. DIF statistics are often difficult to interpret. They are used to make decisions in the controversial and emotionally charged contexts of item and test bias. Furthermore, the decisions associated with DIF are likely to be scrutinized in the adversarial arenas of legislation and litigation.
