Title: Applications of Bayesian item response theory
Authors: Chelsea Parlett - Chapman University (United States) [presenting]
Erik Linstead - Chapman University (United States)
Susanne Jaeggi - University of California Irvine (United States)
Grace Lin - Massachusetts Institute of Technology (United States)
Abstract: Item Response Theory (IRT) models are common in psychometrics, but can have applications to many other fields. The nature of IRT leads to a large number of model parameters (at minimum one parameter per subject and one per item), which can be difficult to estimate in smaller datasets where IRT may otherwise be useful. In addition to the typical benefits of Bayesian models, additional information and/or precision from priors can be helpful in fitting IRT models with many parameters. We will explore Bayesian IRT models using the language Stan, and extend the IRT framework to Beta Regression using Stan. Applications to behavioral data will also be shown.