The effects of persons, items, and subtests on UH ELIPT reading test scores

Abstract

The purpose of this study is to investigate the test reliability of the reading test part of the University of Hawaii English Language Institute Placement Test (UH ELI Placement Test) using both classical theory and G-theory. Furthermore, it will investigate the error variance of this reading test as well as how to make the test more efficient and increase its reliability through further rational modifications, such as changing (increasing or decreasing) the number of subtests and/or items. To those ends, this paper will address the following research questions: 1. What do the classical testing analyses show us about the reliability of the subtests and the overall reading test of the ELI Placement Test? 2. What are the G-study variance components for persons, items, and subtests, and their interactions in the test? Zhang - Effects of Persons, Items, and Subtests on UH ELIPT Reading Test Scores 109 3. What are the D-study dependability estimates for various numbers of items and subtests?