Markov chain Monte Carlo Estimation Practice in Psychometric Models

Stochastic estimation processes known as Markov chain Monte Carlo
(MCMC) methods have drawn an increasing amount of attention in statistics
within the last twenty years.  MCMC provides a virtually universal tool to
deal with integration (and optimization) problems, and this estimation
method gives us the possibility of solving some of the challenges associated
with the traditional gradient-based maximum likelihood (ML) method.  The
main purpose of this presentation is to introduce MCMC estimation and
illustrate its practice to a variety of psychometric models, including
structural equation modeling (SEM) and item response theoretic (IRT) models.