Professor Carol Woods

Department of Psychology, Washington University

Title: Davidian-Curve Item Response Theory

Abstract: Following a brief introduction to unidimensional item response models and the marginal maximum likelihood approach to fitting them, a new method called Davidian-curve item response theory (DC-IRT) will be introduced. In DC-IRT, the latent density is estimated simultaneously with the item parameters as a "seminonparametric" density function called a Davidian curve. Simulation results show that when the latent distribution is skewed, DC-IRT results improve upon those obtained under the usual assumption that the latent density is normal.