MCMC implementations are programmed in Visual Fortran 8.0 (IMSL 5.0) and Dynamic Link Libraries (dll) are made. The dll's are accessible in Splus 6 and R 2.2.1 or higher. Software for a specific model contains of a function that controls the input and a function that controls the output. The output is always stored in one object. This object is defined as a list and contains all input and output data. The software enables the estimation of multilevel IRT models for binary and polytomous data, randomized item response models, and other models will follow.
The software is meant for non-commercial purposes. There is a help function within the MLIRT R-package. Examples will be provided. The software is tested but I cannot guarantee that there are no bugs.
I developed a set of functions for R and Splus that support the main routines: R and S. I have constructed a R-package (microsoft windows) for multilevel IRT modeling: mlirt 1.0. This includes examples and help functions. The package is easily installed by selecting in the R-menu "Packages" and "Install package(s) from local zip file(s)".
Splus 6 Functions
Multilevel IRT for polytomous and/or randomized response data: Mlirt-grm. This contains the dll, the input and output functions. Note that in function EstMLIRTGRM the path to GRM.dll needs to be specified twice. In Splus the symbol \\ separates directories.
Two-parameter IRT model for binary and randomized response data: 2PM. Again the path to 2PL.dll needs to be specified in estIRT2P.ssc.
The Splus 6 software with a dialog in S-Plus to handle the input and specify the model is based on Fortran 6.1. In the document manual a detailed description is given how to install the program and use it in a proper way. Remarks are given about what the program can do or will not do.
Multilevel IRT manual and software:
A zip file containing the Fortran computing code in S.dll
S-Plus 6 code for handling the Fortran code BasicMlirt
Data regarding the example as described in the manual and Fox & Glas (2001): Example
R Funtions:
A R-package (Version R 2.6.0) for multilevel IRT modeling: mlirt 1.0.
Multilevel IRT for polytomous and/or randomized response data: Mlirt-grm. Note that in function EstMLIRTGRM the path to MLIRTGRM-R.dll needs to be specified twice.
Two-parameter IRT model for binary and randomized response data: 2PM. Again the path to 2PL-R.dll needs to be specified in estIRT2P.R.