A METHOD OF ASSESSING BIAS IN TEST ITEMS

This paper presents a method for assessing bias in test items. In this procedure an unbiased item is defined as an item for which the probability of a correct response is the same for any person of a given ability level regardless of that person's ethnic group membership. Total raw score on a homogeneous subtest or subgroup of items is used as an ability measure, with the distribution divided into some number of ability subgroups, each defined by a range of total scores. A study of bias conducted during the item analysis of the "Metropolitan Readiness Tests" is used to illustrate the results which can be obtained using this procedure. The procedure is contrasted with certain other methods used for the assessment of bias in test items, and evidence for validity is presented. Issues arising from the use of total score as an ability measure are also discussed.