Files
Abstract
The goal of this dissertation is to investigate the measurement invariance related to affective variables used in international educational research. Large-scale test assessments, such as PISA, require invariance in order to compare across countries. One approach to examining invariance is based on residual analyses that examine the differences between the observed data and the expected values based on a measurement model. In addition to examining item fit, this study stresses the description of methods for evaluating person fit for affective scales including proposed substantive interpretations for person misfit. Data from four countries (6,856 from Chile, 6,351 from Japan, 4,848 from Turkey, 4,978 from United States) are used to illustrate the framework. The mathematics self-efficacy, anxiety, self-concept, and behavior scale were used, and students responded in four categories. This dissertation explored model-data fit of both items and persons evaluated from the perspective of Rasch measurement theory using data from PISA 2012.