EDITOR'S NOTE: Beginning with this issue, JEM will publish a separate section containing comments on
Frank B. Baker · 2016
Scheuneman (1979) recently proposed a technique for assessing bias in test items. The method basically attempts to use a statistic similar to X2 to index the goodness-of-fit of the proportion of correct response functions to an item yielded by two racial groups. The basic data was obtained from a grouping of total test scores into J mutually exclusive and exhaustive intervals. Within each test score interval, the proportion of examinees in each racial group which answered the item correctly was observed. The resulting data for each group is analogous to an item characteristic curve except that test score has replaced ability as the criterion variable creating a pseudo item characteristic curve.