The issue of differential item functioning (DIF) in standardized tests has increasingly generated interest in the measurement and testing communities. An item is said to contain DIF if examinees of equal proficiency from different gender, ethnic or other groups have an unequal probability of responding correctly to the item. / Although the majority of DIF research has focused on its identification through statistical procedures, recent published studies have addressed the arguably more important issue of causes of DIF. To date, however, most studies of causes of DIF have been concerned with post-secondary situations. / The present study identified sources of DIF within a widely used secondary school achievement battery. Responses on the Vocabulary and Reading Comprehension sections of the Stanford Achievement Test were obtained from 1580 White and 3223 Hispanic eighth graders in Dade County, Florida. A quantitative technique was used to detect items exhibiting DIF. Once these items were identified, a review panel of expert bilingual judges examined them in terms of linguistic and cultural factors associated to DIF between Hispanics and Whites. / Results suggest that, when comparing Hispanic and White students of the same ability, the use in test items of true cognate words frequently used in Spanish will favor the performance of Hispanics. In contrast, several conditions may favor Whites, including the use of non true cognate words infrequently used in English, linguistic complexity, idiomatic expressions derived from technical language, and poetry. Moreover, words and phrases with a special cultural meaning for one of the groups will favor members of that group (White or Hispanic) and, the use of settings for which Hispanic students are likely to be less familiar with will favor the performance of Whites. / Findings from the study confirm the need to stress certain areas in the instruction of Hispanic students. Even though DIF is not necessarily indicative of item bias, its appearance in test items might be a sign of instructional deficiencies. Findings also should alert test developers to distinguish between construct-relevant DIF and DIF associated with test invalidity. / Source: Dissertation Abstracts International, Volume: 54-09, Section: A, page: 3411. / Major Professor: Albert C. Oosterhof. / Thesis (Ph.D.)--The Florida State University, 1993.
| Identifer | oai:union.ndltd.org:fsu.edu/oai:fsu.digital.flvc.org:fsu_77007 |
| Contributors | Montero, Eiliana., Florida State University |
| Source Sets | Florida State University |
| Language | English |
| Detected Language | English |
| Type | Text |
| Format | 172 p. |
| Rights | On campus use only. |
| Relation | Dissertation Abstracts International |
Page generated in 0.0138 seconds