The Split-Sample Design: Studying Effects of Providing Accommodations

The diagram below shows the sampling plan used to study the effect of permitting assessment accommodations.

	Split-sample design The national sample was split. In part of the schools, accommodations were not permitted for students with disabilities (SD) and students classified as limited-English proficient (LEP) or as English language learners (ELL) (ELL is the term used by NAEP for the 2005 assessment and beyond). In the other schools, accommodations were permitted for SD and/or LEP (or ELL) students who routinely received them in their school assessments.
	Accommodations-not-permitted results The accommodations-not-permitted results include the performance of students from both samples who were not classified as SD and/or LEP (ELL) and the performance of SD and/or LEP (ELL) students from the sample in which no accommodations were permitted.
	Accommodations-permitted results The accommodations-permitted results also include the performance of students from both samples who were not classified as SD and/or LEP (ELL); however, the SD and/or LEP (ELL) students whose performance is included in this set of results were from the sample in which accommodations were permitted. Since students who required testing accommodations could be assessed and represented in the overall results, it was anticipated that these results would include more special-needs students and reflect a more inclusive sample.

SOURCE: U.S. Department of Education, Office of Educational Research and Improvement, National Center for Education Statistics, National Assessment of Educational Progress (NAEP), 2001 Geography Assessment.