Abstract
The purpose of this study was to provide an empirical test of the widely held belief that performance-based assessment provides a fairer picture of English-language learners' mathematical skills and knowledge than does a standardized assessment. Specifically, I compared the performance of 94 third-graders on the measurement subscale of the Florida Comprehensive Assessment Test (FCAT) mathematics test to their performance on a set of measurement reasoning and applications that was drawn from their third-grade hands-on science curriculum. Then, I present examples within the non-standardized testing setting where students were provided with real-time language-based accommodations as recommended by the research literature. Finally, I looked at how well these students' level of English language proficiency predicted performance on each of the two assessments. English proficiency level failed to predict FCAT performance. It did predict performance on the reasoning and applications tasks. These findings present a challenge to the conventional wisdom that performance-based assessments provide a less-biased picture of ELL's mathematical knowledge than do standardized tests.