Lee, D., & Tindal, G. (2000). Differential item functioning (DIF) as a function of test accommodation . University of Delaware Education Research and Development Center.

Attachment study from the Inclusive Comprehensie Assessment Systems Project

Broken Link: http://www.doe.state.de.us/aab/DSTP_research.html

The same math word problem items were administered to students in two different testing conditions, the standard paper-and-pencil test and the video read-aloud version.

A total of 483 grade 4 students and 269 grade 5 students from eight states (Alabama, Delaware, Louisiana, North Carolina, Oregon, Rhode Island, Texas, and Wyoming) participated. Seventy-two percent of the sample was White, 16.8% Black, 7% Hispanic, and 2.5% Asian/Pacific. The majority of students were students in regular education. Students in special education and students receiving Title I support comprised 18% and 17% of the sample respectively. Among the students in special education, the majority were classified as having learning disabilities.

Examiners selected 60 problems from a multiple-choice math test from one of the participating states. Two alternate forms were created by matching items, objectives, and proportion correct.

On one form of the test administered to 4th graders, half of the wordy items were flagged as displaying statistically significant DIF (80% of these favored the video testing condition). Two thirds of the wordy items favored the video testing condition. On the other form of the test administered to 4th graders, half of the wordy items were also flagged as displaying DIF (60% of these favored the video testing condition). Only a small number of the wordy items were detected as showing DIF by different testing conditions from 5th grade data (one item from the first form and two items from the second form). However, other statistical data showed that the majority of the wordy items favored the video testing condition.