Date of Award
Doctor of Philosophy
Multinomial processing tree (MPT) models, as a family of hierarchical multinomial models tailored by cognitive theories, have been proven to be successful and applied to cognitive psychometrics Traditional MPT models measure the probability of success for each cognitive stage given their hierarchical relationships. However, this measure neither addresses individual and item difference, nor characterizes the subject's ability and the difficulty of the cognitive stage. In this study, I extend the cognitive stage parameter in MPT models to a Rasch model, and recruit MPT models for a source monitoring paradigm as an example to demonstrate the extension. To evaluate the properties of Rasch MPT models, I conduct systematic simulation studies to test parameter recovery under different conditions including various sample sizes, boundary values of parameters, and missing data. In addition, I use a simple lexical decision experiment and a set of force concept inventory (FCI) multiple-choice questions which are a popular measurement tool in physics teaching research to demonstrate and validate the practical uses of Rasch MPT modeling.
dissertation or thesis originally submitted to the local University of Memphis Electronic Theses & dissertation (ETD) Repository.
Tang, Quan, "Extending MPT Models to Rasch MPT: A General Framework, Demonstration, and Applications" (2013). Electronic Theses and Dissertations. 822.