Randall, C. A. (1999). Score comparability of standard and nonstandard administrations of a state-mandated fifth grade science assessment (Publication No. 9946118) [Doctoral dissertation, University of Kansas]. ProQuest Dissertations and Theses Global. https://www.proquest.com/docview/304520600

University of Kansas (Lawrence, KS); ProQuest document ID: 304520600

The oral delivery ("read aloud") accommodation was investigated.

Extant data from one of the test forms (Form 52) for grade 5 students from throughout Kansas (U.S.), totaled 17,233 participant scores. Form 52 was selected for analysis because of the larger number of students with learning disabilities (LD) receiving accommodations during the test. Students with disabilities numbered 1,348; students with LD numbered 844. One-hundred twenty-five students with LD received a reading accommodation, and 718 did not. Participants' demographic characteristics, including sex and ethnicity, were also reported.

Extant data from the 1997 Kansas Science Assessment selected response items, which tested data interpretation skills, knowledge/ experimental design, and integrated process skills, were analyzed for reliability, item functioning, and factor structure.

The scores of students without disabilities and students with LD receiving accommodations and students with LD not receiving accommodations were all not significantly different from one another. It was determined that the relatively few item differences were not both statistically and practically significant (that is, scored differently and more difficult for one group than another). Factor analysis found that the test measured only a general science factor for all three student groups, indicating that test scores had the same meaning whether students used the accommodation or not. Limitations of the study were reported, and future research possibilities were suggested.