Item Illustration Complexity and the Performance of English Language Learners in a Science Test

• In Event: Roundtable Session 61
  In Roundtable Session: Large-Scale Assessments for Students With Disabilities and English Language Learners: Test Design and Student Characteristics

Abstract

We address the need for enhanced practices that contribute to more valid and fair testing for English language learners (ELLs). We examine the formal properties of illustrations used in a science test and examine how those properties influence student performance. We tested ELLs and non-ELLs with 27 vignette-illustrated items whose illustrations were created systematically according to a procedure developed for such purpose. We will code the properties of the illustrations by using our coding system for examining characteristics of illustrations in science items (Author, 2011a, 2011b) and will measure their complexity. We will use regression analysis to examine the impact of item illustration complexity on student performance in science.

Authors

• Chao Wang, University of Colorado - Boulder

• Magda Yanira Chia, University of Colorado - Boulder

• Rachel R. Kachchaf, TERC

• Guillermo Solano-Flores, University of Colorado - Boulder