Validating Questionnaire Constructs in International Studies: Two Examples from PISA 2000.

One of the most salient requirements for international educational research is the use of comparable measures. For the comparison of student performance across countries the use of item response theory (IRT) scaling techniques facilitates the collection of cross-nationally comparable measures. But there is also a need for valid and comparable context variables, such as family background, learning context, motivational factors, and so on. A wide range of student and school information was gathered through the Program for International Student Assessment (PISA) student and school questionnaires. Most of the theoretical constructs were measured through sets of items that needed to be validated across countries. After the process of construct validation, IRT modeling was used to scale the items. This method not only provides a more sophisticated scaling technique, but also yields an elegant way of dealing with incomplete data (missing values). This paper describes the process of cross-country validation in the PISA study with two examples and discusses its limitations and problems. An appendix contains three tables of PISA data. (Contains 3 graphs, 10 tables, and 11 references.) (Author/SLD)