Abstract:
Construct equivalence is one of the critical concepts in cross-cultural/lingual study. A construct is a theoretical representation of the underlying trait, concept, attribute, process, and/or structure that the test is designed to measure (Messick, 1989). Loosely speaking, construct equivalence implies that persons with the same underlying attribute/construct have the same expected true score at the item level, the sub scale total score level, or both (Raju, Laffitte, & Byrne, 2002; van de Vijier&Tanzer, 1997). Without the establishment of construct equivalence, it is difficult to meaningfully interpret observed score difference across groups. Using Program for International Student Assessment (PISA) 2006, this paper evaluates construct equivalence in terms of measurement invariance in the framework of item response theory (IRT). |
