Date of Award
Morgridge College of Education, Educational Leadership and Policy Studies
Ellen Steiner, Ph.D.
English language learners, Plain language, Universal design
Colorado assesses approximately 98% of students enrolled in grades 3-10 in the area of mathematics. In 2005, 69,872 English Language Learners (ELLs) in grades 3-10 participated in the state content area assessments. In 2007, 88,060 ELLs participated in the state content area assessments. With this dramatic growth in the ELL population in the state of Colorado and elsewhere in the nation, ensuring that ELLs have access to comprehensible materials on state assessments is of paramount importance. Accommodations provide access for students based on their individual needs. Likewise, assessment items designed and/or revised using the principles of plain language and universal design increase access to the content area information in assessments. Colorado’s work in plain language revision of Colorado Student Assessment Program (CSAP) items has been extensive.
Using statewide CSAP data for 2005 and 2007, this study examined how mathematics assessment items, once revised for plain language, performed compared to the original, unrevised version of the item. Access for English Language Learners was a focal point for this study. Plain language revision was evaluated for its effects on item difficulty, item discrimination, DIF and DOK.
DOK ratings from 2005 and 2007 were compared using Chi-sqaure. One sample t-tests were used to examine the differences in DIF and item difficulty (p-value), as well as item discrimination for ELLs and Non-ELLs comparing the 2005 non-revised items with the 2007 revised items. Unrevised items were used as controls. Revised items were classified by type of linguistic change. Each type of change was examined using ANOVA to determine if a specific type of change effected differences in DIF and item difficulty.
While results indicate little if any difference in item performance at individual grade levels, overall LEP performance on revised items was higher than on non-revised items, a change in performance not present for non-ELLs. This may indicate that plain language revision continues to hold promise as a best practice for item development. As an accommodation, however, it appears that plain language revision alone may not be sufficient to ensure access to assessment items for ELLs.
Copyright is held by the author. User is responsible for all copyright compliance.
Holly E. Baker
Received from ProQuest
Baker, Holly E., "Effects of Plain Language Revision on Item Difficulty, Discrimination and DIF" (2008). Electronic Theses and Dissertations. 48.
Educational tests & measurements