Both MML-EM and MHRM algorithms were implemented using the statistical software R (R Core Team, 2015) and version 1.
However, using priors with the MHRM algorithm increased estimation time up to only 12 seconds in the small sample side, and up to 22 in the large sample size.
Results show that in simple models there were no real differences between the MML-EM estimation implemented by default in the mirt packages and MHRM, except for a few seconds of computation time.
13 of the mirt package for MML-EM and MHRM (Chalmers, 2012), and rjags for MCMC by Gibbs-sampling (Plummer, 2015).
Thus, we applied MML-EM and MHRM using mirt with a uniform prior for [SIGMA], and MCMC using rjags with the inverse-Wishart prior.
lower absolute bias and RMSE) with uninformative priors when the sample size was small for the MCMC and MHRM algorithms.
In general, the MCMC and MHRM algorithms rendered similar results, whereas MML-EM provided less precise estimates.
Looking closer at the MCMC and the MHRM results, the MCMC solution offered a slightly less biased estimation than the one reached with the MHRM algorithm.
The MHRM was the fastest algorithm, converging on average in 32-58 seconds when using informative priors, and converging in 33-52 seconds with uninformative priors.
Besides this, the MHRM method--an improvement of MML-EM designed to overcome those situations where computational complexity of the Gauss-Hermite numerical integration procedure involved in MML-EM is cumbersome--achieved its objective of reducing computational time while providing precise estimates.
We have used the same chain length and burn-in period values for the MHRM and MCMC as in the previous studies.
Azad noted that the MHRM
at ADU consists of 8 courses (24 credit hours) of coursework which covers a variety of Central Issues in Human Resources Management, Organization Theory and Design, International Human Resource Management, Compensation and Employment Relations, Human Resources Productivity, Human Resources Strategy, as well as, some fundamental topics in management.