THE RELATIONSHIP OF ITEM BIAS IN STANDARDIZED TESTS TO INSTRUCTIONAL DEFICITS IN EDUCATION AMONG WHITE AND BLACK STUDENTS

The present study compared the test performance of white and black students in order to examine the hypothesis that observed ethnic group differences in item performance on standardized achievement tests might occur as a result of instructional deficits in education among black students. The population examined in this study consisted of white and black examinees tested during the 1977-78 standardization of the Metropolitan Achievement Test (MAT). Two homogeneous subtests of the MAT, Reading and Language, were examined. First, the stability of item-bias indices computed by two item-bias methods (transformed item-difficulty and modified chi-square methods) were examined across two test administrations. Second, the agreement between the item-bias indices computed by both item-bias methods was investigated at each test administration. After the characteristics of the item-bias indices were examined, the study investigated the question of whether the items that were identified as biased against black relative to white students would be similar to the items that showed performance differences between the younger students relative to the older students within the same ethnic group. Significant relationships were generally obtained between the two sets of item-bias indices computed by each method for two successive administrations of the Reading and Language tests. This relationship between item-bias indices indicated that the indices were stable over test administrations. In general, moderate but statistically significant relationships were obtained between the sets of item-bias indices computed by the two item-bias methods at each test administration, indicating that the two methods tended to identify the same items as biased at a given test administration. Significant relationships were generally not obtained between items which produced performance differences between white and black students at a given age and items which produced performance differences between younger and older subjects within the white group and within the black group, though the results were more encouraging using the within black group. Suggestions for future research were presented