Monte Carlo estimation of the conditional rash model (Q1297847)

The Markov chain Monte Carlo (MCMC) method developed by \textit{C. J. Geyer} and \textit{E. A. Thompson} [J. Roy. Stat. Soc. Ser. B 54, No. 3, 657-699 (1992)] is applied to the conditional estimation of person parameters in the Rasch model which is a probabilistic model for intelligence and attained tests. The MCMC scheme provides the alternative to exact calculations of normalizing denominators in a likelihood and its results can be very accurate. The detail testing of MCMC algorithm has been carried out and results of its comparison with two algorithms (Liou's extended algorithm and sum algorithm) for exact calculations are presented.