Standardized Achievement Tests and English Language Learners: Psychometrics Issues

Using existing data from several locations across the U.S., this study examined the impact of students’ language background on the outcome of achievement tests. The results of the analyses indicated that students’ assessment results might be confounded by their language background variables. English language learners (ELLs) generally perform lower than non-ELL students on reading, science, and math–a strong indication of the impact of English language proficiency on assessment. Moreover, the level of impact of language proficiency on assessment of ELL students is greater in the content areas with higher language demand. For example, analyses showed that ELL and non-ELL students had the greatest performance differences in the language-related subscales of tests in areas such as reading. The gap between the performance of ELL and non-ELL students was smaller in science and virtually nonexistent in the math computation subscale, where language presumably has the least impact on item comprehension. The results of our analyses also indicated that test item responses by ELL students, particularly ELL students at the lower end of the English proficiency spectrum, suffered from low reliability. That is, the language background of students may add another dimension to the assessment outcome that may be a source of measurement error in the assessment for English language learners. Further, the correlation between standardized achievement test scores and external criterion measures was significantly larger for the non-ELL students than for the ELL students. Analyses of the structural relationships between individual items and between items and the total test scores showed a major difference between ELL and non-ELL students. Structural models for ELL students demonstrated lower statistical EDUCATIONAL ASSESSMENT, 8(3), 231–257 Copyright © 2002, Lawrence Erlbaum Associates, Inc.