Koretz, D., & Hamilton, L. (1999). Assessing students with disabilities in Kentucky: The effects of accommodations, format, and subject (CSE Report No. 498). University of California, Los Angeles, National Center for Research on Evaluation, Standards, and Student Testing (CRESST). https://cresst.org/publications/cresst-publication-2870/
Multiple accommodations were studied, including those allowed through the Kentucky Instructional Results Informational System (KIRIS; 1997). Common accommodations offered include oral presentation (read-aloud), paraphrasing, dictating answers, cueing, technological aid, and sign interpreter. Students commonly received more than one accommodation.
The extant dataset from the statewide assessments in Kentucky (U.S.) was analyzed; therefore, participants included all students with and without disabilities participating the KIRIS. These included students in grades 4, 5, 7, 8, & 11. Nearly half of these students had specific learning disabilities. More than a fourth had mild intellectual disabilities. Others included those with speech/language disabilities (especially in the earlier grades), students with emotional/behavioral disabilities, and visual, hearing, and physical/orthopedic disabilities.
Performance on the KIRIS (reading, math, science, and social studies subject areas) was used as the dependent variable. This included both multiple choice and open response items.
Analyses indicated that accommodation effects were strongest on open-response items. The differences in scores between disabled and non-disabled students tended to be larger on multiple-choice components in the early grades. This study was intended to follow-up a study of earlier test data (Koretz, 1997). In the current analysis, it was suggested that frequency of accommodation use was similar to that in the analysis of previous data; however, the mean performance of students receiving accommodations was lower in the second analysis. The dictation accommodation appeared to have a lessened impact on scores in this second set of data than in the earlier data set. The authors suggest that the results from this second analysis (lower scores) seem more plausible. [See also Koretz & Hamilton (1999); Koretz & Hamilton (2000)]