An updated version of this package has been placed on CRAN. The following are
new/fixed in 1.6:
1) The classical() function now computes design-consistent standard errors for a
one-stage cluster sample with unbalanced data.
2) The jml() function now uses better starting values to reduce iterations
needed to converge.
3) A bug in stringMatch() was resolved.
4) Fixed spelling error in print.summary.cheat function ('cheatering' to
'cheating')
5) Added in mml() function and methods. The mml() function maximizes the
marginal likelihood function to obtain estimates of the population parameters
(i.e., mean and variance of a normal distribution) treating item responses and
item parameters as known. Function currently works for 1-, 2-, and 3PL IRT
models.
6) The package vignette is updated to reflect new technical methods and
examples.
Recommendations, bug fixes, reactions are always welcome.
Harold
[[alternative HTML version deleted]]