Studying the Equivalence of Two Language Versions of a Large Scale Assessment: A Comparison of Test Takers in United States and Puerto Rico

This study collected evidence regarding the equivalence of items across language forms by applying DIF methodologies to both the cognitive and affective domains of a large scale assessment and explored the utility of the Liu-Agresti estimator of the cumulative common odds ratio for identifying polytomous DIF. The illustrated use of the Liu-Agresti estimator appears to be promising to the understanding of the phenomenon of polytomous DIF. Although the differential functioning of the polytomous items identified as large DIF could not be accounted for based on translation, it does not follow that there are not other causes for the apparent differential functioning. In particular, since these items tap into affect, behavior and attitudes, it could be cultural differences or impact that account for such differences.