Nonparametric Bayesian Modeling of Item Response Curves With a Three-Parameter Logistic Prior Mean
Item response theory (IRT) models the relationship between a person’s observed performance on a test, and a latent trait or ability the test is seeking to measure. IRT is a wide and actively researched ﬁeld. In this work, we restrict our attention to the case of binary responses where items on a test are coded simply as right or wrong, and also make the assumption that the tests we are dealing with are unidimensional, measuring just one trait. The relationship between values of the trait and the probability of a correct or “1” response is modeled with the item characteristic curve (ICC).