Assessment of Bayesian Item Response Models
The underlying assumptions of Bayesian item response models have to be examined to ensure their credibility and that meaningful inferences can be made. A set of tools will be discussed for testing model assumptions and hypotheses. This set of tools includes methods based on Bayesian residuals and predictive diagnostic checks. It will be shown that related computations can be done during an MCMC estimation procedure or afterwards using MCMC output.
KeywordsOutlying Probability Local Independence Item Response Model Grade Response Model Posterior Predictive Distribution
Unable to display preview. Download preview PDF.