ERIC Number: ED486238
Record Type: Non-Journal
Publication Date: 2005-Jul
Analyzing Results of Large-Scale Assessments to Ensure Universal Design. Technical Report 41
Johnstone, Christopher J.; Thompson, Sandra J.; Moen, Ross E.; Bolt, Sara; Kato, Kentaro
National Center on Educational Outcomes
Universal design of assessment has been an important step forward in making tests more accessible to students with disabilities. An issue affecting the universal design approach is the need to review individual items, potentially hundreds of them. Ideally, there would be a statistical procedure available that would first identify items that are potential sources of problems for students with disabilities. This paper illustrates one method of determining whether items are functioning differentially for students with disabilities in comparison to their non-disabled counterparts. Using various statistical analysis techniques, a large statewide mathematics data set was investigated for items that may have design issues. Multiple methods were used as a means for compensating for the lack of statistical power that is often present when analyzing data for populations with small group sizes. Results indicated that items can be flagged for further review based on differential functioning across disability groups or types of analysis.
Descriptors: Statistical Analysis, Disabilities, Measurement Techniques, Data Analysis, Test Items, Student Evaluation, Evaluation Methods, Item Analysis, Grade 4, Grade 8
National Center on Educational Outcomes, University of Minnesota, 350 Elliott Hall, 75 East River Road, Minneapolis, MN 55455. Tel: 612-624-8561, Fax: 612-624-0879; Web site: http://nceo.info.
Publication Type: Reports - Evaluative
Education Level: Grade 4; Grade 8
Authoring Institution: National Center on Educational Outcomes, Minneapolis, MN.