Cho, H. J., Lee, J., & Kingston, N. (2012). Examining the effectiveness of test accommodation using DIF and a mixture IRT model . Applied Measurement in Education , 25 (4), 281–304. https://doi.org/10.1080/08957347.2012.714682

Extant data were selectively analyzed from students with disabilities using a combination or package of three accommodations: read-aloud, frequent breaks, and separate quiet setting.

State (U.S.) assessment scores from all test-takers in grades 3–8 were examined for the item response theory (IRT) modeling. Student data were analyzed further to identify matched pairs for differential item functioning (DIF) comparisons of accommodated students with disabilities and non-accommodated students without disabilities. Four disability categories—learning disabilities, other health impairments, speech/language disabilities, and emotional disabilities—composed the majority (at least 65%) of the students with disabilities group in each grade level. Other demographic data were reported.

This study examined extant state mathematics assessment data. The content included algebra, data, geometry, and number and computation. Detailed item characteristics were reported by the research team.

In this investigation of validity of assessment accommodations, item functioning was examined. Item types included story, explanation, and straightforward, and story items were found sometimes not to be most difficult, challenging a common expectation that items involving reading would be the most difficult. DIF items that favored students with disabilities were neither more nor less difficult than those that favored students without disabilities, particularly for grades 4, 6, 7, and 8. The item types and features that were related to item difficulty varied by grade level. For instance, grade 5 story items with calculations were more difficult than grade 5 story items without calculations, but grade 7 story items without calculation were more difficult than arithmetic items with 1-step calculations. DIF analysis indicated that 101 of the 470 items functioned differently for accommodated students with disabilities and non-accommodated students without disabilities. DIF was present in items at each grade level across item content, with about 36% having uniform DIF. About 34% of DIF items benefited accommodated students with disabilities and 66% benefited non-accommodated students without disabilities; the same proportion of story items with DIF favored each group. Limitations of the study were reported, and future research directions were suggested.