Bayesian modelling of the abilities in dichotomous IRT models via regression with missing values in the covariates (Q2330488)

Often the ability of applicants is deduced from the responses to a contextual questionnaire. The usual theory behind is Item Response Theory (IRT). In this paper the authors investigate the case of responses to the questionnaire with missing values. Particularly for a dichotomous 3 Parameter Normal Ogive (3PNO) model they propose a fully Bayesian approach which includes the 3PNO model on the first level, the linear regression to explain the abilities on the second level and the missing structure on the third one. They use the MCMC algorithm to explore the posteriori distribution. Simulated and real examples are presented.